Q6T46nT668w6i3m

The bad actor might be the model itself, e.g., returning unwanted pornography or violence. Do you have a problem with Google’s SafeSearch?

What are your ideas?

The parent is asking about weights and from that perspective, _stochastic approximation_ is essential.

Riot Games is a subsidiary of Tencent.

Ultimately, algebra _is_ important and calculus is a useful setting to practice algebraic manipulation. Nevertheless, the trend in modern mathematics education in the United States is increasingly towards a more balanced understanding (e.g., greater emphasis on geometric interpretations).

The obvious “next step” is experiential, i.e., write complicated programs. Books about writing compilers, databases, operating systems, emulators, file systems, etc. are all useful.

I’m personally excited by this paradigm. A few years back I had success using a similar architecture for polynomial root finding. I think it’s entirely possible to be really ambitious and reverse engineer new and useful generalized functions.

I don’t care about their sandbox or distributed system. They are irrelevant to the method. The missing language model for program generation is disappointing but I imagine anyone interested in replication, myself included, would prefer to roll their own.

I regularly reference Seminumerical Algorithms and Combinatorial Algorithms.

It isn’t. It’s “mostly deterministic,” whatever that means.

https://platform.openai.com/docs/guides/text-generation/repr...

As someone who has written many compilers and type checkers, I agree. There’s very little information online about the subject. I think part of the issue, for me, was psychological: I felt like I was missing some implementation theory that was presented alongside other compiler subjects (e.g., LALR) when the reality is that you’re mostly implementing very simple Boolean operations.

Any? I regularly find bugs. I don’t find every bug, but finding some bugs is better than none. Especially for the minimal cost of writing annotations.

Awesome. This would be especially useful in science but the lack of GPUs is a non-starter. :(

I’d think you’d want to know dot and norm before embeddings. Especially since most people learn mathematical objects from the perspective of the operators that construct and manipulate them.

Neat idea! Are the batches encoded as tokens into the input sequence? This is something I really like about the multi-modal PALM papers since it enables the multi-modal tokens to be referenced.

I like the PlayDate but the display is horrible.

This type of code isn’t executed by the Python interpreter.

Yep. I’ve been writing a molecular dynamics simulator in PyTorch and the reduction in scope is wild because of heterogenous operators and automatic differentiation.

Maker of popular game assets.

Toy? SymPy has room for improvement but it has made a tremendous impact in research and industry.

It’s entirely reasonable to desire boundaries between nothing and … the universal approximation theorem!

That’s science. You can’t expect everyone to know everything. It’s a preprint so this is the first opportunity to provide feedback.

You’re comparing a subjective claim to an objective claim. Let’s be clear, transformers do not “understand.”

It’s a preprint on arXiv, not the Magna Carta.

There’s been numerous articles in nearly every field of science about this!

It’s true that Fortran played a fundamental role in the area of numerical linear algebra but most of the linear algebra packages you’re likely using today were written in C or C++.

Why is Fortran relevant? Because of that aforementioned history. However, Fortran’s use in numerical software has steadily been declining (e.g., look at the past two decades of TOMS articles).

You identified its purpose: symmetry.

Episodes of This Old House.

Yeah. I feel the same. But I also feel the same about mode collapse! It’s often useful for debugging issues that might otherwise go undetected.

I didn’t look at the code but it looks like they are using the Knuth-Plass algorithm.