About me

I am a research scientist at OpenAI, working on interpretability of language models, for AI safety. Before that, I was a postdoc at UC Berkeley in the Gallant Lab. I defended my Ph.D. in 2018, in the Image, Data, Signal department at Telecom ParisTech in France, supervised by Alexandre Gramfort and Yves Grenier. I graduated from Ecole polytechnique in 2013 and EPFL in 2015. More details can be found in my resume.

My work focuses on developing machine learning and signal processing methods, for interpreting human/animal brain recordings (electrocorticography, magnetoencephalography, functional magnetic resonance imaging, neuron spikes) and silicon brain recordings (large language model activations). I have also been a core developer of scikit-learn between 2015 and 2022.

News

Blog posts

Publications

2024

Leo Gao, Tom Dupré la Tour, Henk Tillman, Gabriel Goh, Rajan Troll, Alec Radford, Ilya Sutskever, Jan Leike, Jeffrey Wu
arXiv preprint, 2024
[pdf] [arxiv] [blog] [code]
Catherine Chen, Tom Dupré la Tour, Jack L. Gallant, Dan Klein, Fatma Deniz
Communications Biology, 2024
[pdf] [biorxiv] [bibtex]

2023

Emily X. Meschke*, Matteo Visconti di Oleggio Castello*, Tom Dupré la Tour, Jack L. Gallant
bioRxiv preprint, 2023
[pdf] [biorxiv] [bibtex]
Fatma Deniz, Christine Tseng, Leila Wehbe, Tom Dupré la Tour, Jack L. Gallant
Journal of Neuroscience, 2023
[pdf] [biorxiv] [bibtex]

2022

Thomas Moreau, Mathurin Massias, Alexandre Gramfort, Pierre Ablin, Pierre-Antoine Bannier, Benjamin Charlier, Mathieu Dagréou, Tom Dupré la Tour, Ghislain Durif, Cassio F. Dantas, Quentin Klopfenstein, Johan Larsson, En Lai, Tanguy Lefort, Benoit Malézieux, Badr Moufad, Binh T. Nguyen, Alain Rakotomamonjy, Zaccharie Ramzi, Joseph Salmon, Samuel Vaiter
NeurIPS, 2022
[pdf] [arxiv] [bibtex] [poster] [code]
Tom Dupré la Tour, Michael Eickenberg, Anwar O. Nunez-Elizalde, Jack L. Gallant
NeuroImage, 2022
[pdf] [biorxiv] [bibtex] [code]

2021

Tom Dupré la Tour, Michael Lu, Michael Eickenberg, Jack L. Gallant
NeurIPS workshop SVRHM, 2021
[pdf] [bibtex]

2019

Laetitia Grabot, Tadeusz W. Kononowicz, Tom Dupré la Tour, Alexandre Gramfort, Valérie Doyère, Virginie van Wassenhove
Journal of Neuroscience, 2019
[pdf] [bibtex]

2018

Tom Dupré la Tour
PhD Thesis, 2018
PhD student award - 1st prize at Université Paris-Saclay STIC doctoral school [link]
PhD thesis award - 1st prize in Signal, Image and Vision, from the Club EEA, GRETSI and GdR ISIS [link] [news]
[pdf] [book_pdf] [tel] [pastel] [bibtex] [slides]
Tom Dupré la Tour*, Thomas Moreau*, Mainak Jas, Alexandre Gramfort
NeurIPS, 2018
[pdf] [arxiv] [bibtex] [poster] [code]
Tom Dupré la Tour, Yves Grenier, Alexandre Gramfort
ICASSP, 2018
[pdf] [bibtex] [poster]

2017

Tom Dupré la Tour, Lucille Tallot, Laetitia Grabot, Valérie Doyere, Virginie van Wassenhove, Yves Grenier, Alexandre Gramfort
PLOS Computational biology, 2017
[pdf] [biorxiv] [bibtex] [poster] [slides] [code]
Mainak Jas*, Tom Dupré la Tour*, Umut Şimşekli, Alexandre Gramfort
NeurIPS, 2017
[pdf] [arxiv] [bibtex] [poster] [code]
Tom Dupré la Tour, Yves Grenier, Alexandre Gramfort
ICASSP, 2017
[pdf] [bibtex] [poster] [code]

Talks

2021

with Fatma Deniz, Jack L. Gallant, and Matteo Visconti di Oleggio Castello
CCN Keynote and Tutorial, 2021, online
[tutorial] [video]
Noninvasive Mathematics, 2021, Genoa (Italy)
[slides] [video]
with Matteo Visconti di Oleggio Castello
Cognitive Neuroscience Colloquium, 2021, Berkeley (USA)
[slides]

2018

2017

Coupling and Causality in Complex Systems (C3S), 2017, Cologne (Germany)
[slides]
Organization for Human Brain Mapping (OHBM), 2017, Geneva (Switzerland)
[poster]
International Biomedical and Astronomical Signal Processing (BASP) Frontiers workshop, 2017, Villars-sur-Ollon (Switzerland)
[poster]

2016

PyParis, 2016, Paris (France)
[slides]

Software

Machine Learning
General machine learning package.
Python - Core developer

Phase amplitude coupling
Estimation of PAC in neural time series. Implementation of driven auto-regressive (DAR) models.
Python - Main developer

Convolutional sparse coding
Convolutional sparse coding, univariate and multivariate models, Gaussian and alpha-stable noise distributions.
Python - Core developer

Multiple-target linear models
Regularized linear models, computational efficiency for large numbers of targets, CPU/GPU.
Python - Main developer

Tutorials for the voxelwise modeling framework
Voxelwise modeling is a framework to perform functional magnetic resonance imaging (fMRI) data analysis.
Python - Main developer