Carles Sáez

HILO DE TEOREMAS DE TOPOLOGÍA ALGEBRAICA Un calendario de adviento con teoremas de topología algebraica e hilos de comentarios de cada teorema

Given what current-gen LLMs (say, in math, but whatever) can do, I think their apparent limitations are kind of mysterious. What is the blocker preventing, at present, high quality fully autonomous work?

I did not know that, for any random variable x, | mean(x) - median(x) | <= stdev(x) Direct proof:

A few months ago I bumped into Anand Patel, who had been my algebraic geometry TA in college, visiting Google DeepMind. He agreed to try out an agent I was building called Aletheia. Fast forward: Anand prompted Aletheia to solve a problem about simplicity of the Hodge bundle on M_g that had been floating around (a part of) the algebraic geometry community for at least ten years. Check out his paper at arxiv.org/pdf/2603.19052

academic conferences are crazy because it’s like Let’s go to the most beautiful place in the world and then look at powerpoints for 12 straight hours

The newest model in the Mamba series is finally here 🐍 Hybrid models have become increasingly popular, raising the importance of designing the next generation of linear models. We've introduced several SSM-centric ideas to significantly increase Mamba-2's modeling capabilities without compromising on speed. The resulting Mamba-3 model has noticeable performance gains over the most popular previous linear models (such as Mamba-2 and Gated DeltaNet) at all sizes. This is the first Mamba that was student led: all credit to @aakash_lahoti @kevinyli_ @_berlinchen @caitWW9, and of course @tri_dao!

Mamba 3 is out! Excited to see where this goes. Two years ago we put out U-Mamba — applied SSMs to biomedical image segmentation (CT, MRI, endoscopy, microscopy) for the first time! We never submitted it to a journal. Just dropped it on arXiv. 1000+ citations later, I think that was the right call. The idea was simple: long-range dependency is the bottleneck in medical imaging, not just NLP. SSMs fix that without the quadratic cost of attention. Turns out that's true across modalities. Mamba 3 doubling inference throughput at half the state size is good news for anyone trying to deploy this in clinical settings — volumetric scans aren't going to get smaller. Paper: arxiv.org/abs/2401.04722 Code: github.com/bowang-lab/U-M…

At last someone saying it!! After going to the theatre and seeing the movie I was very confused about all the hype, when PTA has waaaaay better films

Comenzamos a presentar los #retosrsme de este #idm314 con @juliomulero de la @UA_Universidad abriendo puertas hacia las matemáticas ¿Darás con la solución? CC @UAdivulga

how to math

How often do LLMs claim to prove false mathematical statements? In our latest benchmark, BrokenArXiv, we find they do so very often. The best model, GPT-5.4, only rejects 40% of incorrect statements obtained by perturbing recent ArXiv papers, and other models do much worse.

@jos3ant0n2 En mis tiempos con una FPI me pagaban 16k 🫠 (También es verdad que fue antes del covid y la inflación que hubo los siguientes años)

Por si a alguien le interesa hacer un doctorado en cosas aplicadas bien pagado

@AcerFur We live in very dark times...

Insane to be treating the death of people’s lives like some game score.

1) what

Hard to believe this comes from an official account. The frivolity of the White House wrt war is astounding

I can't believe I hadn't heard this quote before

Qué les pasa a los físicos con las tablas de Clebsh-Gordan?

@saberiguanyar Abans: - Hi havia més matemàtics a tuiter - Preguntaves coses més fàcils

@TonyTheLion2500 More like ruins your life 🫠

So having a PhD in mathematics doesn't solve your life?