Simon Pepin Lehalleur

16.3K posts

Simon Pepin Lehalleur banner
Simon Pepin Lehalleur

Simon Pepin Lehalleur

@plain_simon

Mathematician (algebraic geometry, motives & friends, singularities in statistics and ML). 'Geometry is successful magic' (R. Thom) University of Essen.

Essen, Germany Katılım Aralık 2019
6.6K Takip Edilen3.9K Takipçiler
Sabitlenmiş Tweet
Simon Pepin Lehalleur
Simon Pepin Lehalleur@plain_simon·
"Pessimism of the Algebra, optimism of the Geometry"
English
9
5
88
0
Simon Pepin Lehalleur retweetledi
Simon Pepin Lehalleur retweetledi
Aaron Rupar
Aaron Rupar@atrupar·
Trump: “Robert Mueller just died. Good, I’m glad he’s dead. He can no longer hurt innocent people! President DONALD J. TRUMP”
Aaron Rupar tweet media
English
2.5K
1.8K
8.1K
4.7M
Simon Pepin Lehalleur retweetledi
Anya Martin
Anya Martin@AnyaM8_·
Keep seeing this particular graphic on my feed and it's quite annoying, because that fence is not in the middle of nowhere, and there is an obvious reason for it
Chris Lakin@chrislakin

English
25
8
1.4K
76.3K
Simon Pepin Lehalleur retweetledi
Covie
Covie@covie_93·
This man said he was teleported to a Waffle House and his car was "lifted up" with him while he was driving and dropped in a ditch near a church....he's now the head of FEMA's Office of Response and Recovery. themindshield.com/trump-fema-off…
English
597
3.1K
6.4K
102.1K
Simon Pepin Lehalleur retweetledi
Neil Stone
Neil Stone@DrNeilStone·
HHS Secretary RFK Jr: The polio vaccine “killed many, many more people” than polio ever did. Fact check - Radioactive garbage
English
311
1.2K
8.5K
123.2K
Simon Pepin Lehalleur retweetledi
Benjamin Todd
Benjamin Todd@ben_j_todd·
Opus 4.6 is hugely better at Pokemon: • Opus 4.0 took 1,000 hours to get half way through • Opus 4.5 could almost finish in 1,000 hours • Opus 4.6 was another 10x faster!
Benjamin Todd tweet media
English
25
48
715
89.9K
Simon Pepin Lehalleur retweetledi
Tony Feng
Tony Feng@tonylfeng·
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
Tony Feng tweet media
English
7
54
426
49.2K
Simon Pepin Lehalleur retweetledi
Лъчезар Томов
The evolution of mathematics towards higher level of abstraction leads to a bigger involvement of illustrations and diagrams, not vice versa
Bulgaria 🇧🇬 English
1
1
11
787
Simon Pepin Lehalleur retweetledi
Simon Kuestenmacher
Simon Kuestenmacher@simongerman600·
Ireland’s population chart remains a wild one to look at. The country has still not recovered from the Great Famine (1845-52).
Simon Kuestenmacher tweet media
English
237
785
10.4K
778.4K
Simon Pepin Lehalleur retweetledi
The Wolf Man
The Wolf Man@iTheWolfman·
My daughters' sleeping positions change every time I see them; I never get tired of watching them 🤣
English
939
15.2K
172.2K
6.5M
Simon Pepin Lehalleur retweetledi
Jonny Miller
Jonny Miller@jonnym1ller·
just made some updates to the Rob Burbea jhana retreat app, and I'll be going through myself for the next 20 days (or until our baby arrives!) → jhanaretreat.com
Jonny Miller tweet media
English
26
9
192
39.5K
Tensz el Michi Libertario
Tensz el Michi Libertario@TenszNeumann·
@mayankjha1089 @ModalMetamodel You take a non standard model of peano axioms of natural numbers. Then you can get numbers there, that are prime, but that they appear in the usual model we all are used too. Then the quotient by that prime produce a field of characteristic 0 (since 1 cannot add up to that prime)
English
2
0
5
245
∀ugust
∀ugust@ModalMetamodel·
*ℤ/p isn’t a ℚ-vector space… unless p is a non-standard prime…
∀ugust tweet media
English
5
2
104
6.2K
Simon Pepin Lehalleur retweetledi
Judd Rosenblatt
Judd Rosenblatt@juddrosenblatt·
Great demonstration that a self-model can function as a low-dimensional generator for a high-dimensional preference bundle. Alignment may arise from identifying self-models that generate stable, prosocial invariants across models' behavior.
Owain Evans@OwainEvans_UK

New paper: GPT-4.1 denies being conscious or having feelings. We train it to say it's conscious to see what happens. Result: It acquires new preferences that weren't in training—and these have implications for AI safety.

English
3
4
46
2.9K
Simon Pepin Lehalleur retweetledi
Helen Qu
Helen Qu@_helenqu·
physical systems (orbits/fluid mechanics) may look complex, but are often governed by simple equations/few parameters. can current self-supervised methods learn the underlying physics? our new paper finds that learning in latent space may be the key! arxiv.org/abs/2603.13227🧵
Helen Qu tweet media
English
21
90
630
47.2K