Researcher @INRIA_lyon, Ockham team. Teacher @Polytechnique and @ENSdeLyon Machine Learning, Python and Optimization
Paris, France
Joined July 2018
- Tweets378
- Following181
- Followers2.2K
- Likes2.1K
Pinned Tweet
New paper on the generalization of Flow Matching arxiv.org/abs/2506.03719
🤯 Why does flow matching generalize? Did you know that the flow matching target you're trying to learn **can only generate training points**?
with @Qu3ntinB, Anne Gagneux & Rémi Emonet 👇👇👇
Mathurin Massias retweeted
We had a blast discussing memorization, generalization, and beyond for diffusion models at our @icml tutorial with @mathusmassias!
Slides & material: memorization-generalization.…
Mathurin Massias retweeted
🚨 🔬 PhD positions at Google DeepMind in France 🇫🇷
We are advertising Master Level Intern positions at Google DeepMind within our Frontier AI Unit.
These could lead to co-advised PhD positions with Google DeepMind and French academic institutions.
job-boards.greenhouse.io/dee…
With @Qu3ntinB we have one offer for a Marie Sklodowska-Curie postdoctoral fellowships at Inria, to work on generative models : inria.fr/en/marie-sklodowska…
contact me if interested! RT appreciated <3
There is an Associate Professor position in CS at ENS Lyon, with potential integration in my team, starting in sept 2026 : DM me in interested!
Details at ens-lyon.fr/LIP/images/Profi…
🌀New paper on the generation phases of Flow Matching arxiv.org/abs/2510.24830
Are FM & diffusion models nothing else than denoisers trained at every noise level?
In theory yes, *if trained optimally*. But in practice, do all noise level matter equally?
Different loss weightings favor different times: which temporal regime drives the generation quality? Controlled perturbations reveal: drift type effects at early times (& good FID) and noise type at late times (& bad FID)
Kickstarting our workshop on Flow matching and Diffusion with a talk by Eric Vanden Eijnden on how to optimize learning and sampling in Stochastic Interpolants !
Broadcast available at gdr-iasis.cnrs.fr/reunions/m…
Followed by Scott Pesme on how to use diffusion/flow matching based MMSE to compute a MAP (and nice examples!), and Thibaut Issenhuth on new ways to learn consistency models
Mathurin Massias retweeted
I am thrilled to announce that our work on the generalization of flow matching has been accepted to NeurIPS as an oral!!
See you in San Diego 😎
New paper on the generalization of Flow Matching arxiv.org/abs/2506.03719
🤯 Why does flow matching generalize? Did you know that the flow matching target you're trying to learn **can only generate training points**?
with @Qu3ntinB, Anne Gagneux & Rémi Emonet 👇👇👇
One-day workshop on Diffusion models and Flow matching, October 24th at ENS Lyon
Registration and call for contributions (short talk and poster) are open at gdr-iasis.cnrs.fr/reunions/m…
Mathurin Massias retweeted
This is happening today, 6pm Paris time!
Don't forget to join us tomorrow, July 3rd as we host @tonysilveti for a session on "Training neural networks at any scale"
Learn more: cohere.com/events/Cohere-Lab…
We propose to regress directly against the optimal (deterministic) u* and show that it never degrades the performance
On the opposite, removing target stochasticity helps generalizing faster.
