What is the point of this?

LLMs are trained to predict the next word in a text. In what way, shape or form does that have anything to do with stock market prediction? Completely ridiculous AI bubble nonsense.

No it isnt. Next word prediction is what humans do to communicate anyway so the criticism isnt valid. Except you do that for your own sentences (if you do it for others its considered rude :) ).

Anyways this criticism is now dated given that modern day LLMs can solve unseen reasoning problems such as those found in the IMO.

It does have something to do with the stock market, since its about making hypotheses and trading based off that. However, I'd agree that making a proper trading AI here would require reasoning based fine tuning for stock market trading actions. Sort of like running GRPO taking market feedback as the reward. the article simply cant do that due to not having access to the underlying model weight.

shhh. We need more of these as counter-parties to improve alpha.

Is it the borrow checker? Normally rust had your back when it comes to memory oopsies. Maybe we need a borrow checker for empathy..

Those ideas feels so old already

A super cool feature are labeled switches: https://codeberg.org/ziglings/exercises/src/branch/main/exer...

They allow for super ergonomical coding of state machines, which is a lot of fun.

Could you please elaborate? Where do you run your apps?

Presumably just directly on whatever system they are using. If you have something that can be plopped on whatever linux distro and run reliably then you've already got what docker is supposed to give you.

Exactly as sibling comment says, e.g. let's pretend the popular httpx cli was my project to deploy and run on the server. With only uv installed, i can:

    uv tool run "httpx[cli] @ git+https://github.com/encode/httpx"

To be clear in this example i'm not pulling a package published on pypi, i'm running the HEAD of that git repo (i could do a branch or tag instead). I could use the "uvx" shortcut instead of "uv tool run". I could specify a specific python version (either one already installed on this OS or choose a dist which uv downloads for me).

This caches the deps in an isolated virtual env for me. It'll only download the deps in the first run.

I'm surprised by their extraordinary claim of 100,000 TPS. That would require extraordinary evidence, especially with contention and hot accounts in mind.

I wonder if capillary effect can be used? Or some other mechanism that trees use?

You’re describing osmosis filtering, so yeah pretty solid idea