cochne

Isn’t it the actual number?

This is an amazing and unexpected result. To be clear though, it does not imply that NP is subexponential: "Note that runtime bounds of the form 2^o(n) do not transfer under general polynomial-time reductions (2^n^o(1) would be needed for this), so our result does not imply that general NP problems admit subexponential time algorithms."

Edit: To explain more, it is guaranteed that you can reduce any NP problem to 3SAT in polynomial time with a polynomial factor in input size. You still have to solve it.

If 3SAT were solvable in polynomial time, then polynomial(polynomial(x)) = polynomial(x) and you're implied to have an overall polynomial time solution.

But what we have here is that 3SAT is solveable in subexponential time. subexponential(polynomial(x)) is not guaranteed to be sub-exponential.

Further edit: The subexponential solution does carry over to any (?) problems that have can be reduced to 3SAT in a linear number of variables. They list: Counting independent sets, Counting proper q-colorings, Counting perfect matchings.

See this from the paper: "Note that runtime bounds of the form 2^o(n) do not transfer under general polynomial-time reductions (2^n^o(1) would be needed for this), so our result does not imply that general NP problems admit subexponential time algorithms."

It doesn't imply NP is sub-exponential, only that 3SAT is. Conversions to other problems can still end up exponential.

>With BERT, you need a large, labeled dataset, and you have to train/fine-tune the model.

I think they were responding to this. You can use BERT to provide zero shot classification predictions.

Are you sure it wouldn’t suggest such a thing? Stoic thinkers (such as the one your repo is named after) are not necessarily anti-suicide. And I think if you prompted a real stoic with the question and the right context, it would certainly suggest it without safe guards.

The public “funded” these models in the sense of contributing to their training data.

Wouldn’t it be 1/(number of days since 1888) for a given day? Or 1/(number of years) for a given year? So less than 1/100 if static and independent

I started using an AI calorie tracking app. For non-branded foods, the calorie counting is pretty inaccurate and sometimes very off. However, having an app that can track pictures and create short descriptions that someone else (my trainer) can easily review has been pretty helpful.

I'm in your boat. I tried out TikTok out a few times, including making a new account, but it never showed me good content. I had maybe one or two longer sessions, but never felt the need to go back, like I (unfortunately) do with Reddit or Youtube. I could never understand why it was so popular, but maybe I'm just a curmudgeon.

There is no thresholding, you delete edges randomly based on the probability assigned.

>It gives me hope that teenagers are watching his videos and becoming inspired to go into infrastructure.

Sadly, that was me ~10 years ago, but the lure of FAANG money was too strong and I went into EE/CS after 1 year as a civil engineering major. I wonder if one day we will really start feeling the affects of this talent reallocation, and civil engineering will become a higher paying profession.

I think they are correct, do you have a source? From my knowledge the only other components are the fully connected networks which are not big contributors.

I hardly think it's fair to say that they were 'taking advantage'. If Google says unlimited, should the typical person really expect that to be taken away? That's a really bad look for Google. "If we offer something that's good value to you, expect it to be taken away suddenly in the future." Those are not the actions of a company I would want to rely on.

I never really got how proofs are supposed to solve this issue. I think that would just move the bugs from the code into the proof definition. Your code may do what the proof says, but how do you know what the proof says is what you actually want to happen?