lebellig
@lebellig.bsky.social
Ph.D. student on generative models and domain adaptation for Earth observation 🛰
Previously intern @SonyCSL, @Ircam, @Inria
🌎 Personal website: https://lebellig.github.io/
Previously intern @SonyCSL, @Ircam, @Inria
🌎 Personal website: https://lebellig.github.io/
Calling it for today... I tried using the gemini 3 Pro preview to build some js animations, and it went well
November 18, 2025 at 8:17 PM
Calling it for today... I tried using the gemini 3 Pro preview to build some js animations, and it went well
Sorry but it's getting pretty when leaving python
November 18, 2025 at 7:46 PM
Sorry but it's getting pretty when leaving python
Interpolation between two gaussian distributions on a flat torus (my personal benchmark for new llms)
November 18, 2025 at 6:43 PM
Interpolation between two gaussian distributions on a flat torus (my personal benchmark for new llms)
"Back to Basics: Let Denoising Generative Models Denoise" by
Tianhong Li & Kaiming He arxiv.org/abs/2511.13720
Diffusion models in pixel-space, without VAE, with clean image prediction = nice generation results. Not a new framework but a nice exploration of the design space of the diffusion models.
Tianhong Li & Kaiming He arxiv.org/abs/2511.13720
Diffusion models in pixel-space, without VAE, with clean image prediction = nice generation results. Not a new framework but a nice exploration of the design space of the diffusion models.
November 18, 2025 at 5:05 PM
"Back to Basics: Let Denoising Generative Models Denoise" by
Tianhong Li & Kaiming He arxiv.org/abs/2511.13720
Diffusion models in pixel-space, without VAE, with clean image prediction = nice generation results. Not a new framework but a nice exploration of the design space of the diffusion models.
Tianhong Li & Kaiming He arxiv.org/abs/2511.13720
Diffusion models in pixel-space, without VAE, with clean image prediction = nice generation results. Not a new framework but a nice exploration of the design space of the diffusion models.
"Curly Flow Matching for Learning Non-gradient Field Dynamics" @kpetrovvic.bsky.social et al. arxiv.org/pdf/2510.26645
Solving the Schrödinger bridge pb with a non-zero drift ref. process: learn curved interpolants, apply minibatch OT with the induced metric, learn the mixture of diffusion bridges.
Solving the Schrödinger bridge pb with a non-zero drift ref. process: learn curved interpolants, apply minibatch OT with the induced metric, learn the mixture of diffusion bridges.
November 12, 2025 at 8:09 PM
"Curly Flow Matching for Learning Non-gradient Field Dynamics" @kpetrovvic.bsky.social et al. arxiv.org/pdf/2510.26645
Solving the Schrödinger bridge pb with a non-zero drift ref. process: learn curved interpolants, apply minibatch OT with the induced metric, learn the mixture of diffusion bridges.
Solving the Schrödinger bridge pb with a non-zero drift ref. process: learn curved interpolants, apply minibatch OT with the induced metric, learn the mixture of diffusion bridges.
Great article! But can the preference score go up to 2? You know, because 1 just isn’t aesthetic enough.
October 31, 2025 at 9:37 AM
Great article! But can the preference score go up to 2? You know, because 1 just isn’t aesthetic enough.
“Entropic (Gromov) Wasserstein Flow Matching with GENOT” by D. Klein et al. arxiv.org/abs/2310.09254
Transport between two distributions defined on different spaces by training a noise-to-data flow models in the target space, conditioned on the source data and leveraging Gromov–Wasserstein couplings
Transport between two distributions defined on different spaces by training a noise-to-data flow models in the target space, conditioned on the source data and leveraging Gromov–Wasserstein couplings
October 30, 2025 at 10:43 PM
“Entropic (Gromov) Wasserstein Flow Matching with GENOT” by D. Klein et al. arxiv.org/abs/2310.09254
Transport between two distributions defined on different spaces by training a noise-to-data flow models in the target space, conditioned on the source data and leveraging Gromov–Wasserstein couplings
Transport between two distributions defined on different spaces by training a noise-to-data flow models in the target space, conditioned on the source data and leveraging Gromov–Wasserstein couplings
Diffusion Transformers with Representation Autoencoders by Boyang Zheng, et al (arxiv.org/abs/2510.116...)
Unexpected result: swapping the SD-VAE for a pretrained visual encoder improves FID, challenging the idea that encoders' information compression is not suited for generative modeling!
Unexpected result: swapping the SD-VAE for a pretrained visual encoder improves FID, challenging the idea that encoders' information compression is not suited for generative modeling!
October 14, 2025 at 7:08 PM
Diffusion Transformers with Representation Autoencoders by Boyang Zheng, et al (arxiv.org/abs/2510.116...)
Unexpected result: swapping the SD-VAE for a pretrained visual encoder improves FID, challenging the idea that encoders' information compression is not suited for generative modeling!
Unexpected result: swapping the SD-VAE for a pretrained visual encoder improves FID, challenging the idea that encoders' information compression is not suited for generative modeling!
"How to build a consistency model: Learning flow maps via self-distillation" by @nmboffi.bsky.social et al (arxiv.org/abs/2505.18825)
New method to train flow maps without any pretrained flow matching/diffusion models!
New method to train flow maps without any pretrained flow matching/diffusion models!
October 10, 2025 at 7:15 AM
"How to build a consistency model: Learning flow maps via self-distillation" by @nmboffi.bsky.social et al (arxiv.org/abs/2505.18825)
New method to train flow maps without any pretrained flow matching/diffusion models!
New method to train flow maps without any pretrained flow matching/diffusion models!
"Be Tangential to Manifold: Discovering Riemannian Metric for Diffusion Models" Shinnosuke Saito et al. arxiv.org/abs/2510.05509
High-density regions might not be the most interesting areas to visit. Thus, they define a new Riemannian metric for diffusion models relying on the Jacobian of the score
High-density regions might not be the most interesting areas to visit. Thus, they define a new Riemannian metric for diffusion models relying on the Jacobian of the score
October 8, 2025 at 9:57 AM
"Be Tangential to Manifold: Discovering Riemannian Metric for Diffusion Models" Shinnosuke Saito et al. arxiv.org/abs/2510.05509
High-density regions might not be the most interesting areas to visit. Thus, they define a new Riemannian metric for diffusion models relying on the Jacobian of the score
High-density regions might not be the most interesting areas to visit. Thus, they define a new Riemannian metric for diffusion models relying on the Jacobian of the score
Grateful for the opportunity to speak at tomorrow’s Learning Machines seminar ([email protected]) on generative domain adaptation and geospatial foundation models benchmarking for robust Earth observation 🌍
Join on Sept 11 at 15:00 CET! www.ri.se/en/learningm...
Join on Sept 11 at 15:00 CET! www.ri.se/en/learningm...
September 10, 2025 at 4:32 PM
Grateful for the opportunity to speak at tomorrow’s Learning Machines seminar ([email protected]) on generative domain adaptation and geospatial foundation models benchmarking for robust Earth observation 🌍
Join on Sept 11 at 15:00 CET! www.ri.se/en/learningm...
Join on Sept 11 at 15:00 CET! www.ri.se/en/learningm...
Late to the party but I like the fact that you can use geodesic random walk (like really simulating the random walks) to derive the SDEs necessary for diffusion models on Riemannian manifolds (from arxiv.org/abs/2202.02763)
August 30, 2025 at 9:28 AM
Late to the party but I like the fact that you can use geodesic random walk (like really simulating the random walks) to derive the SDEs necessary for diffusion models on Riemannian manifolds (from arxiv.org/abs/2202.02763)
I'll be at #GRETSI in Strasbourg next week! Friday morning, I'll present our work on Riemannian flow matching for SAR interferometry (generation and denoising) 🛰️
Also really looking forward to the poster sessions and all the exciting conferences on the program!
📄 hal.science/hal-05140421
Also really looking forward to the poster sessions and all the exciting conferences on the program!
📄 hal.science/hal-05140421
August 21, 2025 at 2:34 PM
I'll be at #GRETSI in Strasbourg next week! Friday morning, I'll present our work on Riemannian flow matching for SAR interferometry (generation and denoising) 🛰️
Also really looking forward to the poster sessions and all the exciting conferences on the program!
📄 hal.science/hal-05140421
Also really looking forward to the poster sessions and all the exciting conferences on the program!
📄 hal.science/hal-05140421
Anyone aware of a cats --> pure evil creatures image translation benchmark? not even neural networks’ dreams reached this level of nightmare fuel
August 18, 2025 at 9:51 PM
Anyone aware of a cats --> pure evil creatures image translation benchmark? not even neural networks’ dreams reached this level of nightmare fuel
New episode in this line of work from @giannisdaras.bsky.social et al. on training diffusion models with mostly bad/low-quality/corrupted data (+few high-quality samples). This time for proteins!
📄 Ambient diffusion Omni: arxiv.org/pdf/2506.10038
📄 Ambient Proteins: www.biorxiv.org/content/10.1...
📄 Ambient diffusion Omni: arxiv.org/pdf/2506.10038
📄 Ambient Proteins: www.biorxiv.org/content/10.1...
July 7, 2025 at 7:43 PM
New episode in this line of work from @giannisdaras.bsky.social et al. on training diffusion models with mostly bad/low-quality/corrupted data (+few high-quality samples). This time for proteins!
📄 Ambient diffusion Omni: arxiv.org/pdf/2506.10038
📄 Ambient Proteins: www.biorxiv.org/content/10.1...
📄 Ambient diffusion Omni: arxiv.org/pdf/2506.10038
📄 Ambient Proteins: www.biorxiv.org/content/10.1...
Added to my reading list: Adjoint Schrödinger Bridge Sampler by Guan-Horng Liu et al. arxiv.org/abs/2506.22565
July 1, 2025 at 7:11 PM
Added to my reading list: Adjoint Schrödinger Bridge Sampler by Guan-Horng Liu et al. arxiv.org/abs/2506.22565
Drop the "conditional", just "flow matching", it's cleaner.
June 24, 2025 at 3:24 PM
Drop the "conditional", just "flow matching", it's cleaner.
I was intrigued by "Mean Flows for One-Step Generative Modeling" and, in particular, how it handles averaging the marginal velocity field during training. In practice, they don't and replace it with the conditional one in their loss function. I wonder how mismatches impact generation...
June 11, 2025 at 7:37 AM
I was intrigued by "Mean Flows for One-Step Generative Modeling" and, in particular, how it handles averaging the marginal velocity field during training. In practice, they don't and replace it with the conditional one in their loss function. I wonder how mismatches impact generation...
"Energy Matching: Unifying Flow Matching and
Energy-Based Models for Generative Modeling" by Michal Balcerak et al. arxiv.org/abs/2504.10612
I'm not sure EBM will beat flow-matching/diffusion models, but this article is very refreshing.
Energy-Based Models for Generative Modeling" by Michal Balcerak et al. arxiv.org/abs/2504.10612
I'm not sure EBM will beat flow-matching/diffusion models, but this article is very refreshing.
May 21, 2025 at 12:37 PM
"Energy Matching: Unifying Flow Matching and
Energy-Based Models for Generative Modeling" by Michal Balcerak et al. arxiv.org/abs/2504.10612
I'm not sure EBM will beat flow-matching/diffusion models, but this article is very refreshing.
Energy-Based Models for Generative Modeling" by Michal Balcerak et al. arxiv.org/abs/2504.10612
I'm not sure EBM will beat flow-matching/diffusion models, but this article is very refreshing.
"Probability Density Geodesics in Image Diffusion Latent Space" by Qingtao Yu et al. https://arxiv.org/abs/2504.06675
They propose an algorithm to traverse high-density regions when interpolating between two points in a diffusion model latent space.
They propose an algorithm to traverse high-density regions when interpolating between two points in a diffusion model latent space.
April 29, 2025 at 5:02 PM
"Probability Density Geodesics in Image Diffusion Latent Space" by Qingtao Yu et al. https://arxiv.org/abs/2504.06675
They propose an algorithm to traverse high-density regions when interpolating between two points in a diffusion model latent space.
They propose an algorithm to traverse high-density regions when interpolating between two points in a diffusion model latent space.
TerraMind (previous post), the new geospatial generative model from IBM and ESA, was trained on the TerraMesh dataset. Great to see large-scale datasets with aligned modalities! Now waiting for the release :)
📄 https://arxiv.org/abs/2504.11172
📄 https://arxiv.org/abs/2504.11172
April 28, 2025 at 4:02 PM
TerraMind (previous post), the new geospatial generative model from IBM and ESA, was trained on the TerraMesh dataset. Great to see large-scale datasets with aligned modalities! Now waiting for the release :)
📄 https://arxiv.org/abs/2504.11172
📄 https://arxiv.org/abs/2504.11172
IBM and @esa.int introduce TerraMind, a new geospatial generative model. With 'Thinking in Modalities', they generate missing modalities during model finetuning.
I'm curious about its modality translation capabilities 👀
📄 https://arxiv.org/abs/2504.11171
🐍 https://huggingface.co/ibm-esa-geospatial
I'm curious about its modality translation capabilities 👀
📄 https://arxiv.org/abs/2504.11171
🐍 https://huggingface.co/ibm-esa-geospatial
April 28, 2025 at 3:30 PM
IBM and @esa.int introduce TerraMind, a new geospatial generative model. With 'Thinking in Modalities', they generate missing modalities during model finetuning.
I'm curious about its modality translation capabilities 👀
📄 https://arxiv.org/abs/2504.11171
🐍 https://huggingface.co/ibm-esa-geospatial
I'm curious about its modality translation capabilities 👀
📄 https://arxiv.org/abs/2504.11171
🐍 https://huggingface.co/ibm-esa-geospatial
I really liked this approach by @matthieuterris.bsky.social et al.They propose learning a unique lightweight model for multiple inverse problems by conditioning it with the forward operator A. Thanks to self-supervised fine-tuning, it can tackle unseen inverse pb.
📰 https://arxiv.org/abs/2503.08915
📰 https://arxiv.org/abs/2503.08915
April 26, 2025 at 4:02 PM
I really liked this approach by @matthieuterris.bsky.social et al.They propose learning a unique lightweight model for multiple inverse problems by conditioning it with the forward operator A. Thanks to self-supervised fine-tuning, it can tackle unseen inverse pb.
📰 https://arxiv.org/abs/2503.08915
📰 https://arxiv.org/abs/2503.08915
Very cool article from Panagiotis Theodoropoulos et al: https://arxiv.org/abs/2410.14055
Feedback Schrödinger Bridge Matching introduces a new method to improve transfer between two data distributions using only a small number of paired samples!
Feedback Schrödinger Bridge Matching introduces a new method to improve transfer between two data distributions using only a small number of paired samples!
April 25, 2025 at 5:03 PM
Very cool article from Panagiotis Theodoropoulos et al: https://arxiv.org/abs/2410.14055
Feedback Schrödinger Bridge Matching introduces a new method to improve transfer between two data distributions using only a small number of paired samples!
Feedback Schrödinger Bridge Matching introduces a new method to improve transfer between two data distributions using only a small number of paired samples!
It reminds me of this result (from arxiv.org/abs/2104.11222), you can finetune the jpeg compression level of your training set to have the best FID on church images.
March 19, 2025 at 6:28 PM
It reminds me of this result (from arxiv.org/abs/2104.11222), you can finetune the jpeg compression level of your training set to have the best FID on church images.