Viewing profile — rosstaylor90

Many math questions are easy to verify, and it's a classic benchmark for reasoning -> so it's a good hill to climb. I agree with your meta-point that better benchmarks testing more…

RL has more than two steps...

I disagree, 10 years ago AIs nailing these types of competition would have been seen as very impressive. The fact goal posts can move on this now shows how much AI has progressed. …

Thanks for feedback! Yes, we’re looking to improve quality in the coming months. Couple of notes: - The initial use of data is distillation so we’re less bound by question quality …

Happier now, upgraded the backend :) (co-creator here)

What's your AIME 2025 score? https://gr.inc/RJT1990/AIME2025/

No, the primary reason is that the UK is a heavily service-based economy. Quarantine impacts the service-based subset of the economy the most. Therefore it makes sense that the UK …

Ed Thorp’s Princeton/Newport Partners was the first great quant hedge fund.

I'm from Suffolk - my school used to go here on school trips. The funny thing about the site is that it's basically just a hill in a field now, so needless to say it was a massive …

ATLAS ML | London, UK | Full Stack Developer | Remote | FULL-TIME Our mission is to systemise the world's deep learning knowledge. We are building a discovery platform and tooling …

I think your conclusion is too strong. Yes, we know bigger models and data generally lead to better performance (e.g. BigGAN results last year) but progress in architecture can sti…

Philosophical Investigations - Wittgenstein. Just stunningly thoughtful and changed the way I interpret most things around me.

Cool example for a music video here: https://www.youtube.com/watch?v=-k9qDxyxS3s .

So many top articles recently are about a Google product, a startup doing their own “privacy-first” version of a Google product, or someone talking about a Google product. Why the …

Hey all, One problem we face is how unmodular deep learning code can be, meaning that adapting someone else's code for your own dataset, your own workflow, your own way of training…

I'm going to get one printed in real-life - once I do some superresolution and probably a bit of manual Photoshop editing. Which design do you guys like best?