Alexia Jolicoeur-Martineau (@jm

Alexia Jolicoeur-Martineau

9,856 posts

Alexia Jolicoeur-Martineau

@jm_alexia

Principal Researcher @ Microsoft 🐱‍💻 2025 ARC Prize Winner I build generative AI for images, videos, text, tabular data, weights, molecules, and video games.

Montréal, Québec

ajolicoeur.wordpress.com

Joined March 2017

Pinned
Alexia Jolicoeur-Martineau
@jm_alexia
Oct 7, 2025
New paper 📜: Tiny Recursion Model (TRM) is a recursive reasoning approach with a tiny 7M parameters neural network that obtains 45% on ARC-AGI-1 and 8% on ARC-AGI-2, beating most LLMs. Blog: alexiajm.github.io/2025/09/29/tin… Code: github.com/SamsungSAILMon… Paper: arxiv.org/abs/2510.04871
alexiajm.github.io
Less is More: Recursive Reasoning with Tiny Networks
Recursive Reasoning with Tiny Networks
703K
Alexia Jolicoeur-Martineau
@jm_alexia
Sep 14, 2025
I'm working on super tiny neural networks that perform better than the big ones, it's kinda insane.
242K
Alexia Jolicoeur-Martineau
@jm_alexia
Oct 8, 2025
7M model coming to crush all benchmarks
167K
Alexia Jolicoeur-Martineau
@jm_alexia
Oct 29, 2024
Statisticians are still living in the 60's while we now generate videos from text in AI.
Selçuk Korkmaz
@selcukorkmaz
Oct 27, 2024
Rethinking p-Values in Scientific Research This paper addresses the ongoing debate about p-values and offers innovative solutions to improve statistical practice. Let’s explore the key insights! #Statistics #Research #OpenScience #DataScience tandfonline.com/doi/full/10.10…
416K
Alexia Jolicoeur-Martineau
@jm_alexia
Oct 30, 2024
I have been a Statistician for 10 years. Some of the earlier Statisticians invented amazing tools like Random Forest, boosting/bagging, Lasso. Since then I have seen no real progress. The same old debates are ongoing. Statisticians with ambitious goals have all moved to AI.
196K
Alexia Jolicoeur-Martineau
@jm_alexia
Dec 12, 2018
🙀🙀🙀 I just got accepted, I will be starting a PhD at @MILAMontreal next month! 😸 This was my third application, it took two years, but I finally got accepted!
Alexia Jolicoeur-Martineau
@jm_alexia
Oct 9, 2025
To generate 1K tokens, an LLM with 100 layers needs to go through 100*1,000 = 100,000 layers To generate any amount of tokens, TRM do 16 reasoning steps of 2*7*3=42 depth for a total of 16*42 = 672 layers
350K
Alexia Jolicoeur-Martineau
@jm_alexia
Jul 3, 2018
My new paper is out! " The relativistic discriminator: a key element missing from standard GAN" explains how most GANs are missing a key ingredient which makes them so much better and much more stable! #Deeplearning #AI ajolicoeur.wordpress.com/RelativisticGA… arxiv.org/abs/1807.00734
Relativistic GAN
From ajolicoeur.wordpress.com
Alexia Jolicoeur-Martineau
@jm_alexia
Aug 15, 2025
Summary of the findings for the Hierarchical Reasoning Model on ARC
ARC Prize
@arcprize
Aug 15, 2025
Analyzing the Hierarchical Reasoning Model by @makingAGI We verified scores on hidden tasks, ran ablations, and found that performance comes from an unexpected source ARC-AGI Semi Private Scores: * ARC-AGI-1: 32% * ARC-AGI-2: 2% Our 4 findings:
84K
Alexia Jolicoeur-Martineau
@jm_alexia
Feb 19, 2021
If you want to start experimenting with GANs in PyTorch, I have made available the following codes: For GANs on toy data: colab.research.google.com/drive/15LuYU8E… For GANs + denoising score matching hybrids on toy data: github.com/AlexiaJM/Adver… For image-generation: github.com/AlexiaJM/Maxim…
Alexia Jolicoeur-Martineau
@jm_alexia
Jan 31, 2023
I successfully defended my PhD thesis today!
60K
Alexia Jolicoeur-Martineau
@jm_alexia
Sep 14, 2025
Researcher in 2026 >Use LLMs to generate 20 papers >Spend 2 weeks clicking approve/reject >Bill: 20k$ in API credits >Spend 2 weeks verifying them, 1 is good >Publish paper >GitHub issues: Data leakage, bugs; ignore >post on X: Human coding is obsolete! AI invented new science🤯
Aran Komatsuzaki
@arankomatsuzaki
Sep 9, 2025
Google presents an AI system to write expert-level scientific software. Using LLMs + tree search, it invented novel methods in bioinformatics, epidemiology, geospatial analysis & more, often surpassing human SOTA. (1/4)
83K
Alexia Jolicoeur-Martineau
@jm_alexia
Nov 23, 2023
My PhD is finally available online! 😸 Check it out if you are interested in GANs and diffusion models. papyrus.bib.umontreal.ca/xmlui/handle/1…
102K
Alexia Jolicoeur-Martineau
@jm_alexia
Oct 16, 2019
My new paper is out! We show a framework in which we can both derive #SVMs and gradient penalized #GANs! We also show how to make better gradient penalties! ajolicoeur.wordpress.com/MaximumMarginG… arxiv.org/abs/1910.06922
ajolicoeur.wordpress.com
Connections between SVMs, Wasserstein distance and GANs
Check out my new paper entitled “Support Vector Machines, Wasserstein’s distance and gradient-penalty GANs are connected”! 😸 In this paper, we explain how one can derive SVMs and …