@web3celebrity@EricJorgenson@elonmusk It's more like "Engineers sometimes think like this, often with poor mental models that leads to vibes rather than clarity"
@doodlestein@jarredsumner@doodlestein Does your approach has a problem when it's about parallel experimentation? For example, when you do hyper parameter tuning of a model, you might want to change the same code file that contains model config. Wouldn't your approach disallow parallel experiments?
@RichardSSutton But labs are equipping agents with Evaluation already e.g., I'm sure that coding models are trained on vast amount of RL environments without direct human feedbacks. The environment might return execution time of the program or the correctness on some unit tests.
@CSProfKGD@ylecun Predicting outcome correctly is indeed useful. So the question is more about which parts of the outcome matter more than others. For example, predicting that an object will fall matters more than predicting the rotation of the object while it's falling.
@mitchellh Yes, but you also have to understand the theory of constraints (ToC). Is your time best spent in optimizing further? Or is it better spent thinking about something else that matters for your long-horizon goal?
@manicalmechanic@MoneyPlatters@Jackkk Also by classifying him as weird, you have demonstrated that you have no grounded philosophy of life. Because you cannot articulate why it's weird. And I'm sure that if you expose your position, others can easily rebut it. So no respect needed to be given to you here bro.
@manicalmechanic@MoneyPlatters@Jackkk BTW this idea of shifting the goal post to the idea of Bryan being "weird" is a defensive mechanism, because you are not staying in the realm of logic anymore so it's not making sense for people to respond you with logic when you yourself has abandoned logic.
With the Cursor SDK, you can build your own agents with Composer 2.5. It's now available in Python and TypeScript.
This long weekend, Composer usage is 90% off in the SDK. We're excited to see what you build!
Introducing HRM-Text.
An ultra-lean 1B-parameter reasoning language model designed to deliver strong general performance with a fraction of the data, compute, and infrastructure.
Trained on just 40B structured tokens, HRM-Text achieves competitive performance while using ~1/1000 of the training data of comparable models.
The kicker? The full model trains in roughly one day on a $1,000 budget.
This opens the door to a new generation of AI that is powerful, accessible, and radically easier to adapt. Theories and research concepts once deemed too expensive to test are officially back in the game.
Sapient Intelligence invites you to help us shape a new paradigm for general intelligence.
@kcosr The main problem is because of the lack of consistency of the design system. The reason it's hard to read beautiful pages is because their styles vary too much. This means you cannot learn the style and reuse it.
@EXM7777 there are also cognitive load to it as well i.e., every HTML could have different design system or obscure text. this is extraneous load that is not intrinsic to the complexity of the write up making it hard to read the HTML.