eapriv

??? She’s 82.

Are there any pages of French in “Crime and Punishment”?

> Late last year I started writing a math blog and decided to use LLMs to polish/enhance my writing.

Why would you do that in the first place? If you are just starting writing about math, I assume your goal would be to get better at it, and using LLMs is not how you get better at writing.

Looks like AI slop in (some of) the thumbnail images. Why would anyone do that?

The problem is the use of AI. It’s a reliable indicator that the author doesn’t actually care about the quality of the work, so I shouldn’t bother to read the text.

Not sure what “your own” in the title is supposed to mean if you are running a model that you didn’t train using a framework that you didn’t write on a server that you don’t own.

The problem is “to add two numbers”. The meta-problem is “to learn how computers work”.

Removing layers usually improves stability.

Sounds like most of these problems come from using Python.

Spoiler: it’s about AI.

No, it says nothing about LLM output being invertible.

You are wrong. Prime numbers are fundamental to mathematics.

If I were to write such a text, it would have a lot more about building intuition for advanced mathematical concepts. This intuition is extremely valuable, but missing from almost all advanced-level texts. On the other hand, it’s very difficult to put into words, and probably quite personal.

I submitted it, and the word “basic” is mine, because the author doesn’t really go deep into what I would consider “advanced” mathematics. It can be a good prerequisite for advanced things, though.

If you have a third-party dependency that survived for 20 years. But what if you are trying to choose what to rely on today, and to decide if it will even exist in 20 years? Certainly none of the fashionable JavaScript frameworks will.

Isn’t this the whole shtick of Mr Martin, author of “Clean Code”?

You can’t, because a Fourier series is not a linear combination.

It’s not “an uncomputable number”.

Spoiler: it’s not about how GPUs work, it’s about how to use them for machine learning computations.

um, “Earth’s”?

> It wouldn't really be interesting

Andrej Karpathy did exactly that, and I think it’s quite interesting.

I find it hilarious that “from scratch” now somehow means “in PyTorch”.

This is true almost by definition, and doesn’t tell us anything interesting about black holes.

What does it have to do with using LLVM?

Why is it not commutative?

How does it make it less horrible?

Performance of any given CPU instruction is negligible, yet somehow they accumulate to noticeable values.

Great, we can spend crazy amount of computational resources and hand-holding in order to (maybe) reproduce three lines of code.