lebellig @lebellig.bsky.social · 29/09/2026Looking for insights into the performance gap between unconditional and conditional diffusion models when the guidance weight is set to 0 👀 221
lebellig @lebellig.bsky.social · 23/09/2026New interest: super-resolved images from undertrained flow matching models 061
lebellig @lebellig.bsky.social · 10/09/2026So cool that we have mentats to solve all the difficult math problems we can finally spend our free time fighting across sand deserts under a blazing sun. 010
lebellig @lebellig.bsky.social · 09/09/2026Should we index prize money by the compute budget required to solve the same task? 100
Reposted by lebelligELLIS @ellis.eu · 09/09/2026Meet Charlotte Pelletier, Assoc. Prof at Université Bretagne Sud 🇫🇷 & ELLIS Member. She researches AI, particularly in the scope of time series analysis with applications in remote sensing and Earth observation. Her advice for young scientists focuses on investing in a strong professional network. 173
Reposted by lebelligLucas Degeorge @lucasdegeorge.bsky.social · 03/09/2026🚀 New paper: Balancing Frequencies and Pixels in Flow Matching We tackle the low-frequency bias in pixel-space flow matching and train JiT up to 40% faster without any architectural changes. 📄 Read it here: arxiv.org/abs/2609.02748 1248
lebellig @lebellig.bsky.social · 02/09/2026We’ve released the InSARFlow checkpoints, trained on two types of data: 🌎 Real SAR interferograms from Mexico City 🧪 Synthetic dataset + we added a simple notebook to make it easy to get started and experiment with InSARFlow 🧑💻 Joint work with Arnaud Breloy and @nshaud.bsky.social 000
lebellig @lebellig.bsky.social · 02/09/2026We’re excited to release the updated article and code of InSARFlow! 🌊⛵ InSARFlow is a Riemannian flow matching model designed to denoise SAR interferograms while preserving the cyclical nature of phase differences.🌀 📄 Article: hal.science/hal-05710871... 💻 Code: github.com/lebellig/ins... 1113
Reposted by lebelligAl Merose @al.merose.com · 21/08/2026My lab is looking for an Assistant Research Scientist to work on AI for Climate Science. No Ph.D is required for this position, just a BS and relevant experience. The application deadline is Sept 20. Join me in making better climate models! apply.interfolio.com/191796apply.interfolio.com Apply - Interfolio {{$ctrl.$state.data.pageTitle}} - Apply - Interfolio 033
lebellig @lebellig.bsky.social · 19/08/2026Yesterday I watched a movie about an AI4Science researcher who gets angry after having his research grant application rejected (badly explained movie plot). Maybe we should take it as a warning about research funding cuts 241
Reposted by lebelligNeurIPS Europe @neuripseurope.bsky.social · 18/08/2026We are happy to announce that 28 workshops have been accepted for the Paris event, as part of the 102 accepted NeurIPS workshops: blog.neurips.cc/2026/08/10/a... They will take place on Sat Dec 12 + Sun Dec 13, 2026 (for Paris) The suggested deadline for Workshop submissions is close (Aug 29th)!blog.neurips.ccAnnouncing the NeurIPS 2026 Workshops – NeurIPS Blog 0167
Reposted by lebelligClimate AI Nordics @climateainordics.com · 06/08/2026🌱 Big news from Climate AI Nordics (CAIN)! We are officially registered as an NGO! 🌍✨ Network members can now become official members with AGM voting rights to help shape our future. Join us or get involved! More info: climateainordics.com/news/2026-08-04-cain-ngo/ 🤖💚climateainordics.comClimate AI Nordics is now an official NGOClimate AI Nordics becomes an official NGO (ideell förening) 172
Reposted by lebelligAl Merose @al.merose.com · 30/07/2026Here's the tale of how @jder.bsky.social and I scaled Samudra, a neural ocean emulator capable of predicting 8 years of the ocean on a single GPU, to operate at a full 1/4° resolution (16x the size in bytes). It was quite a humbling process. 1183
lebellig @lebellig.bsky.social · 24/07/2026NeurIPS submissions confirmed to be a heat-loving species. Warmer year, bigger bloom. Every degree we add, the deadline gets denser 🌻 Good news for the field, we're having a really good growing season 👨🌾 092
Reposted by lebelligNicolas Audebert @nshaud.bsky.social · 21/07/2026You can train your image-to-image flow matching model on badly aligned data, you just have to tell it how bad it is. 🫣 Great work from @lebellig.bsky.social and Aimi Okabayashi with cool applications to remote sensing. It's FlowEO 2.0! 022
lebellig @lebellig.bsky.social · 21/07/2026We demonstrate the effectiveness of A2BM in real-world super-resolution and EO domain adaptation tasks, including pre/post-flood imagery and historical ↔ RGB sensor translation. Joint work with Aimi Okabayashi, @nshaud.bsky.social, Charlotte Pelletier, Thomas Corpetti and @ncourty.bsky.social! 000
lebellig @lebellig.bsky.social · 21/07/2026We show that increasing the alignment score at inference time improves source fidelity, leading to generated images that better preserve the source content, even when the model is trained on image pairs with low/medium/high alignment (see figure). 100
lebellig @lebellig.bsky.social · 21/07/2026The key idea is to condition the bridge on image-pair alignment, providing a simple and effective way to improve source fidelity at inference time. We explore both metadata-based alignment and embedding-based alignment scores, enabling our method to be applied even when metadata are unavailable 100
lebellig @lebellig.bsky.social · 21/07/2026The last project of my PhD is finally out! 🪴 It was a pleasure collaborating with Aimi on this work! We introduce A²BM: Alignment-Aware Bridge Matching, a new framework for image-to-image translation with weakly aligned image pairs. Paper 📄: arxiv.org/pdf/2607.16294 1144
Reposted by lebelligMathurin Massias @mathurinmassias.bsky.social · 21/07/2026I organize a 1 day workshop on Generative modelling @ENS Lyon, October 9th Call for oral/poster contributions is open; details at gdr-iasis.cnrs.fr/reunions/mod...gdr-iasis.cnrs.frModèles génératifs : diffusion, flow matching - GdR IASISLes demandes de prise en charge de missions par le GdR IASIS doivent parvenir à la gestionnaire du GdR avant le 25 septembre. Les modèles génératifs ont connu de récentes avancées spectaculaires, au p... 0128
Reposted by lebelligAl Merose @al.merose.com · 08/07/2026I'd like to announce that at @openathena.ai, @jder.bsky.social and I helped @m2lines.bsky.social release Samudra 2. We scaled this neural ocean emulator to train on 16x the size of data in bytes on the same hardware budget. We can now skillfully predict 8 years of the ocean on a single GPU at a 1/4°medium.com🌊 Samudra 2: A Fast, Cheap AI Ocean Model, Now at the Scale That MattersM²LInES’ neural ocean emulator now runs multi-year simulations at eddy-permitting resolution on a single GPU, turning a supercomputer-scale… 2606
Reposted by lebelligMathurin Massias @mathurinmassias.bsky.social · 07/07/2026Slides for our ICML tutorial on Memorization and Generalization of Diffusion and Flow Matching Models are now available ! 🌀 memorization-generalization.github.io @quentinbertrand.bsky.socialmemorization-generalization.github.ioICML 2026 Tutorial - Generalization and Memorization in Flow Matching and Diffusion 22212
Reposted by lebelliglebellig @lebellig.bsky.social · 03/07/2026Heading to #ICML2026 🇰🇷 and interested in diffusion models, flow matching, and their generalization capabilities? Don't miss the excellent tutorial by @mathurinmassias.bsky.social and @quentinbertrand.bsky.social on Monday! 📍 Hall D1 🗓️ Monday, July 6 🕘 9:00–11:30 AM Details: icml.cc/virtual/2026... 2116
lebellig @lebellig.bsky.social · 03/07/2026Heading to #ICML2026 🇰🇷 and interested in diffusion models, flow matching, and their generalization capabilities? Don't miss the excellent tutorial by @mathurinmassias.bsky.social and @quentinbertrand.bsky.social on Monday! 📍 Hall D1 🗓️ Monday, July 6 🕘 9:00–11:30 AM Details: icml.cc/virtual/2026... 2116
Reposted by lebelligGuillaume Astruc @gastruc.bsky.social · 23/06/2026🛰️ Introducing UniverSat: one transformer backbone for Earth Observation that handles ANY sensor, ANY spatial, spectral & temporal resolution, ANY scale — with a single set of weights. 🌍 1159
lebellig @lebellig.bsky.social · 22/06/2026I took Lyon's rainy days in early May as a warm welcome, now with the heatwave I think my acclimatisation is complete, so it's time to make it official! I've started a postdoc on generative models in Inria's Ockham team working with @mathurinmassias.bsky.social and @quentinbertrand.bsky.social ☀️ 140
Reposted by lebelligNicolas Dufour @nicolasdufour.bsky.social · 19/06/2026We explored the impact of variability sources in generative modeling. Turns out, we've been neglecting the error bars associated with training variability all along! We should aim to report results that we are sure of their scientific validity, instead of seed engineering! 1174
Reposted by lebelligColeen Murphy @ctmurphy1.bsky.social · 17/06/2026one anecdote: I was searching for a link to one of my old papers on Google, and the automated Gemini summary attributed all of my lab's work to my husband 1041711907
Reposted by lebelligGabriel Peyré @gabrielpeyre.bsky.social · 16/06/2026The alpha version of my new book "Optimal Transport for Machine Learners" is out, with in particular an online version with interactive figures www.gpeyre.com/ot4ml/ 09843
Reposted by lebelligDavid Picard @davidpicard.eurosky.social · 28/05/2026🎆 New paper! "Random Process Flow Matching: Generative Implicit Representations of Multivariate Random Fields", by Julien Lalanne, accepted to ICML'26 🥳 We're proposing flow-matching for inpainting in ultra-sparse setup, with applications to seismic interpolation. 📜 arxiv.org/abs/2605.28625 1/arxiv.orgRandom Process Flow Matching: Generative Implicit Representations of Multivariate Random FieldsGenerative modeling provides a powerful framework for learning data distributions. These models initially relied on probabilistic methods such as Gaussian Processes (GP) for uncertainty-aware predicti... 1324
Reposted by lebelligGilles Louppe @glouppe.bsky.social · 22/05/2026"Accept (spotlight)" at ICML'26 😎 Our paper brings particle filters back to life: autoregressive diffusion models + posterior sampling yield optimal proposals for Bayesian filtering, scaling up to GenCast-sized systems. arxiv.org/abs/2605.20028 w/ Thomas Savary and @francois-rozet.bsky.socialarxiv.orgTraining-Free Bayesian Filtering with Generative EmulatorsBayesian filtering is a well-known problem that aims to estimate plausible states of a dynamical system from observations. Among existing approaches to solve this problem, particle filters are theoret... 03911
Reposted by lebelligNicolas Dufour @nicolasdufour.bsky.social · 20/05/2026Thrilled to share that MIRO is accepted to ICML 2026 @icmlconf.bsky.social ! 🎉 By training on the reward scores, we can simply condition the model on high rewards at inference time to guarantee top-tier, aligned outputs. We’ve updated our paper with some additional results!arxiv.orgMIRO: MultI-Reward cOnditioned pretraining improves T2I quality and efficiencyThe default paradigm of post-training text-to-image generators includes post-hoc selection of generated images, and subsequent training with one reward model to align the generator to the reward, typi... 1366
Reposted by lebelligNicolas Audebert @nshaud.bsky.social · 20/05/2026📢 The TerraBytes workshop is returning for a 2nd edition - this time at ECCV 2026. Submit your paper before June 18th and join in Malmö, Sweden! 🔗 terrabytes-workshop.github.io 074
Reposted by lebelligDavid Picard @davidpicard.eurosky.social · 15/05/2026👏 Folks! If you are curious about the Generative Modeling via Drifting paper, but you find it difficult to understand → I wrote a different interpretation of it. It's called: "An Expectation-Maximization interpretation of Generative Modeling via Drifting" davidpicard.github.io/pdf/An_Expec... 2285
Reposted by lebelliglebellig @lebellig.bsky.social · 28/04/2026My PhD thesis manuscript will be available in the coming months, but I’ve written two blog posts based on the related work chapter: 1. Generative modeling with flow-based models 🪚 2. Data-translation with flow and diffusion bridges 🔨 Open to feedback and discussions! lebellig.github.io/blog/ 1357
Reposted by lebelligAl Merose @al.merose.com · 07/05/2026Want better local models? Want models that are made completely transparently? OpenAthena.ai is hiring on our Marin.community team. We’re especially looking for SWEs and research scientists with experience in data and post training. DM me.openathena.aiOpen AthenaOpen Athena is a nonprofit that accelerates academia with capabilities from the AI frontier 0156
lebellig @lebellig.bsky.social · 07/05/2026I also experimented with approximate SB methods (via minibatch training) on more challenging image translation tasks, e.g. bitmojis → human faces or satellite imagery across different geographic regions, mainly to understand where they fail and explore new bridge-matching methods. 000
lebellig @lebellig.bsky.social · 07/05/2026I’m not an expert on the SB problem, but I’ve seen some applications in geophysical fluid simulation downscaling, single-cell dynamics modeling, and image translation tasks like cats→wild datasets. I think you need the L2 cost to encode the similarity you want between the source and target images. 120
Reposted by lebelligSander Dieleman @sedielem.bsky.social · 06/05/2026My first blog post in over a year is a deep dive on flow maps🗺️, or how to learn the integral of a diffusion model to enable faster sampling and several other cool tricks. It's the longest one yet👀 Let me know what you think! sander.ai/2026/05/06/f...sander.aiLearning the integral of a diffusion modelA deep dive on flow maps. 36417
lebellig @lebellig.bsky.social · 02/05/2026Thanks, Nicolas! It might involve time-dependent vector fields that take points from p0 to p1 👀 010
lebellig @lebellig.bsky.social · 28/04/2026Best of luck for the defense! Will it be livestreamed by any chance? 000
lebellig @lebellig.bsky.social · 28/04/2026My PhD thesis manuscript will be available in the coming months, but I’ve written two blog posts based on the related work chapter: 1. Generative modeling with flow-based models 🪚 2. Data-translation with flow and diffusion bridges 🔨 Open to feedback and discussions! lebellig.github.io/blog/ 1357
lebellig @lebellig.bsky.social · 28/04/2026Delighted to have successfully defended my PhD thesis on "Generative models for Earth Observation, from denoising to domain adaptation" 🪴 This wouldn’t have been possible without the support of my colleagues, my PhD advisor, and my family and friends. Thank you all! 4275