did you write the SEC parsers yourself or use oss/off the shelf tech?

Built it myself. The filing structure is somewhat consistent enough to parse programmatically, then an LLM handles the unstructured parts (with a lot of trial and error)

And you just scrape it straight from sec.gov?

>If instead of a copy-pasting spree, or setting up a whateverClaw, the user might just click a button in Chrome, it could be actually useful. (Consider a dozen such buttons.)

isn't this basically just putting a decision tree on top of the llms?

context switching across the entirety of the feature surface for an app

You could easily have agents to work on login page, messaging feature, database/data model update, recommender system, backend api, etc

We have our doubts about this. Can you share your code or product? Anecdotally, my mistakes and lack of understanding exponentiate the more I try to parallelize.

Who is “we”?

As I said in the neighboring comment, for vibe coding side projects and prototypes for work I just merge and iterate. It works out more than it doesn’t. For anything bigger at work I cannot share as I’m at Apple.

But you have to keep it in your head, and remember all stuff at the same time. How is it possible to track, and do reviews one after another? Or are these pretty long running agents?

I’m not sure what you mean by keep it in your head? I know all of the parts the agents are working on. It’ll often be a mix between bigger tasks (some large refactor, new feature, etc) and small tasks (little bug fixes).

For prototyping I just merge. I don’t bother to review the code. For anything more important than I am reviewing the code and going back and forth. Basically there’s a queue of stuff demanding my attention, and I just serially go through them.

What’s also been really helpful to me is /simplify and similar code review skills (I have my own). That alone takes an agent a while to parse through everything it’s done and self reviews. It catches quite a lot itself this way.

>I’m not sure what you mean by keep it in your head?

If the project I work on is large enough, it takes me some time to get everything I need to understand for review into the short term memory. If it's small enough, it's less of a problem for me.

why not an inbetween scenario like using a managed inference provider to host your own models?

what would be the advantage?

Most orgs don't understand this.

If you've ever been in a meeting with multiple L8's arguing over features, you should be able to estimate how much each hour of that meeting is costing the org.

how do you justify the compute investment for something like nemotron ? especially if all the labs are willing to pay for those same GPU clusters for inference or training runs?

Nemotron has two reasons to exist, both of which are strategic to NVIDIA.

1. Help NVIDIA design future systems for AI by more deeply understanding what it takes to build AI.

2. Keep the AI ecosystem strong and diverse throughout the world by providing AI infrastructure that many companies can innovate on.

This is not a science project, nor is it for the joy of giving something away. Both of these reasons are core to NVIDIA.

Does Nvidia maintain it's own compute hardware expressly for model training? Otherwise, I'm not sure how you keep up with the SOTA model techniques.

Yes

they already pivoted to be the EUROPE ai cloud, this is just the natural consequences of cheerleading that

> One can envision a world in which OpenAI pays chefs money to cook while ChatGPT watches—narrating their thought process, tasting the dishes, and describing the results. This information could be used for general-purpose training, but it might also be packaged as a “book”, “course”, or “partner” someone could ask for.

So we're speed running the idea of AI Facebook friends and creating a new para(ai)social relationship

whatever happened to the system prompt buffer? why did it not work out?

because it's a separate context window, it makes the model bigger, that space is not accessible to the "user". And the "language understanding" basically had to be done twice because it's a separate input to the transformer so you can't just toss a pile of text in there and say "figure it out".

so we are currently in the era of one giant context window.

Also it's not solving the problem at hand, which is that we need a separate "user" and "data" context.

Reminder that Anthropic's goal is to sell you more tokens...

Yes, exactly. That's why their harness has no incentive to help you save tokens. Just conflict of interest.