@vaiter profile picture
Samuel Vaiter @vaiter

@CNRS Researcher

Joined September 2012
  • Tweets 908
  • Following 373
  • Followers 3,070
235 Photos and videos
A quick remark: When you use centered finite difference instead of forward or backward difference, the truncation error decreases faster, but I never realized that the roundoff error grows before!
1
4
18
1,413
Of course you can always use complex step approximation if you don't like roundoff error and you have stable complex arithmetic :) x.com/vaiter/status/18330184…

Samuel Vaiter@vaiter
9 Sep 2024
Complex step approximation is a numerical method to approximate the derivative from a single function evaluation using complex arithmetic. It is some kind of “poor man” automatic differentiation. nhigham.com/2020/10/06/what-…
1
4
261
Samuel Vaiter@vaiter
May 29
Now I am become Death, the destroyer of Quotas.
3
361
Samuel Vaiter@vaiter
May 5
So, yelling at a LLM to simplify a Lean proof really works!
2
5
788
Samuel Vaiter@vaiter
May 4
Now accepted at ICML 2026! Congrats to @MagamedTm
Samuel Vaiter@vaiter
Mar 9
🧵 New preprint! "Towards Understanding Steering Strength" with @MagamedTm and D. Garreau Activation steering is a popular way to control LLM behavior at inference. But how much should you steer? We provide the first theoretical analysis of the steering strength α. 1/7
2
7
56
6,644
🧵 New preprint! "Towards Understanding Steering Strength" with @MagamedTm and D. Garreau Activation steering is a popular way to control LLM behavior at inference. But how much should you steer? We provide the first theoretical analysis of the steering strength α. 1/7
1
19
122
14,802
more replies
We validate everything empirically on 11 models (GPT-2, Gemma 3, Qwen 3, Llama 2, Mistral 7B) across 8 safety-related concepts. All our theorems are confirmed experimentally. 6/7
1
2
590
TL;DR: Steering strength α has precise, predictable effects on LLM outputs. Understanding these laws may help design better steering strategies. Paper: arxiv.org/abs/2602.02712 Code: github.com/MagamedT/steering 7/7

Towards Understanding Steering Strength

A popular approach to post-training control of large language models (LLMs) is the steering of intermediate latent representations. Namely, identify a well-chosen direction depending on the task...

arxiv.org
10
541
Samuel Vaiter@vaiter
15 Dec 2025
Les inscriptions aux journées MODE 2026 à Nice sont désormais ouvertes. Elles se dérouleront du 18 au 20 mars à l'Hôtel Saint-Paul. Les inscriptions sont ouvertes jusqu'au 1 mars (majoration > 9/02). La deadline pour soumettre une communication est le **15 janvier**.
1
1
385
Samuel Vaiter@vaiter
15 Dec 2025
Pour s'inscrire et pour plus d'informations : mode2026.sciencesconf.org Les orateurs et oratrices pléniers sont Pierre Ablin, Yann Brenier, Julie Delon, Stéphane Gaubert, Francisco J. Silva Álvarez (prix J.J Moreau) et Irène Waldspurger. Attention, le nombre de place est limité !
346
Samuel Vaiter@vaiter
3 Dec 2025
I am at #NeurIPS2025, reach out if you want to chat!
5
552
Samuel Vaiter retweeted
Transactions on Machine Learning Research@TmlrOrg
1 Dec 2025
🏅Finalist: Gradient Scarcity in Graph Learning with Bilevel Optimization, by Hashem Ghanem, @vaiter, @n_keriven 5/n openreview.net/forum?id=10YJ…

Gradient Scarcity in Graph Learning with Bilevel Optimization

Gradient scarcity emerges when learning graphs by minimizing a loss on a subset of nodes under the semi-supervised setting. It consists in edges between unlabeled nodes that are far from the...

openreview.net
1
3
11
1,319
Samuel Vaiter retweeted
Fares El Khoury@fares_elkhoury
1 Dec 2025
Happy to be at #NeurIPS2025 in San Diego to present our poster ‘Learning Theory for Kernel Bilevel Optimization’ #3005, Fri at 4:30 p.m. Stop by/ping me to chat, especially about statistics, causality, generative models! Let's connect! Joint w/ E. Pauwels, @vaiter, @MichaelArbel
2
4
814
Samuel Vaiter@vaiter
14 Oct 2025
Les journées SMAI-MODE 2026 auront lieu à Nice du 18 au 20 mars 2026, précédées par un mini-cours les 16/17. Vous pouvez déposer votre proposition de contrib sur mode2026.sciencesconf.org Inscription à partir du 1er décembre, fin le 1 mars (tarif majoré à compter du 1 février).

1
2
1
517
Samuel Vaiter@vaiter
14 Oct 2025
Les exposés pléniers seront donnés par : Pierre Ablin, Julie Delon, Stéphane Gaubert, Irène Waldspurger, ainsi que que le/la lauréat(e) du prix J.J. Moreau. Le mini-cours sera sur les PEP (Performance Estimation Problem) par A. Dieuleveut et A. Taylor
1
365
Happy to share that I have been promoted to senior research scientist (directeur de recherche (*)) at @cnrs effective today! Grateful to my amazing students, collaborators and mentors who made this journey possible. I will continue my research at the math lab of UniCA (LJAD).
4
24
840
(*) directeur de recherche is roughly equivalent to full prof w/o mandatory teaching duty in France. This is an incredible tenured civil servant position, and I am deeply grateful for it. DM me if you have questions about the process to enter CNRS.
1
1
329
At the same time, I cannot write this message without telling that I am **very** worried about the current administrative evolution of our profession which is evolving quickly towards a system absolutely not calibrated to our duties with a purely accounting PoV.
2
298
Load more