flerovium114

Impression 5 rules! My 4-year old loves it, so we have a membership.

Are LLMs doing drug discovery? To my knowledge, that’s all classic “machine learning”

The idea of "pythonizing" resource-constrained environments like embedded seems feasible to me, and even a good idea if executed well (e.g., numpy, torch, jax, etc.), but it does have me cackling imagining the first encounters of modern web devs with performance/power economy constraints

In what setting are you mixing LaTeX and C code?

pizza

Have you tried using Enzyme (https://enzyme.mit.edu/)? It operates on the LLVM IR, so it's available in any language that breaks down into LLVM (e.g., Julia, where I've used it for surface gradients) and it produces highly optimized AD code. Pretty cool stuff.

Randomized numerical linear algebra has proven very useful as well. It allows you to use a black-box function implementing matrix-vector multiplication (MVM) to compute standard decompositions like SVD, QR, etc. Very useful when MVM is O(N log N) or better.