Should we ban Brian Eno from bandcamp?

New blog 💙: I reflect on why I worked on what I worked on... I think a PhD is a very special time. You get to challenge yourself, push your boundaries, and grow. My thoughts go against the current AI/academia narrative online, so I hope you find it interesting. chaitjo.substack.com/p/phd-thesis...

You may add the real test (or training 👀) dataset if you are into leaderboard chasing

1. Select many diffusion/flow-matching models 2. Generate 50k images per model 3. Use FID of each set as a label 4. Train a model to predict FID from a single image What’s the probability this actually works, gives a cheap proxy for FID and enable fast generative model prototyping?

We introduce epiplexity, a new measure of information that provides a foundation for how to select, generate, or transform data for learning systems. We have been working on this for almost 2 years, and I cannot contain my excitement! arxiv.org/abs/2601.03220 1/7

📖 We put together with Mike Davies a review of self-supervised learning for inverse problems, covering the main approaches in the literature with a unified notation and analysis. arxiv.org/abs/2601.03244

Can we train neural networks just with permutations of their initial weights? And then whats the best initialisation distribution ?

PS: We also recently released a unified codebase for discrete diffusion, check it out! 𝕏 Thread : x.com/nkalyanv99/... 🔗 GitHub: github.com/nkalyanv99/... 📚 Docs: nkalyanv99.github.io/UNI-D2/

🆕 “Foundations of Diffusion Models in General State Spaces: A Self-Contained Introduction” Huge thanks to Tobias Hoppe, @k-neklyudov.bsky.social, @alextong.bsky.social, Stefan Bauer and @andreadittadi.bsky.social for their supervision! 🙌 arxiv : arxiv.org/abs/2512.05092 🧵👇

"Improved Mean Flows: On the Challenges of Fastforward Generative Models" arxiv.org/abs/2512.02012 questions this approximation and proposes a new training process for mean flows.

Hi @ google, can you provide 100k TPU hours to explore the design space of diffusion bridges for image-to-image translation? x1 vs drift pred, architectures and # params, # dataset, scaling couplings and batch sizes (for minibatch-based couplings). I can run everything in jax in return...

Yesterday, @nicolasdufour.bsky.social defended is PhD. I really enjoyed the years of collaboration w/ @vickykalogeiton.bsky.social (& @loicland.bsky.social) Video: youtube.com/live/DXQ7FZA... Big thanks to the jury @dlarlus.bsky.social @ptrkprz.bsky.social @gtolias.bsky.social A. Efros & T. Karras

@climateainordics.com is now on youtube! Check out some amazing talks on how to help fight climate change using AI! youtube.com/@climateaino...

@neuripsconf.bsky.social is two weeks away! 📢 Stop missing great workshop speakers just because the workshop wasn’t on your radar. Browse them all in one place: robinhesse.github.io/workshop_spe... (also available for @euripsconf.bsky.social) #NeurIPS #EurIPS

Calling it for today... I tried using the gemini 3 Pro preview to build some js animations, and it went well

Interpolation between two gaussian distributions on a flat torus (my personal benchmark for new llms)

It may not top ImageNet benchmarks, but honestly, that hardly matters... the removal of the VAE component is a huge relief and makes it much easier to apply diffusion models to domain-specific datasets that lack large-scale VAEs.

"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.

We figured out flow matching over states that change dimension. With "Branching Flows", the model decides how big things must be! This works wherever flow matching works, with discrete, continuous, and manifold states. We think this will unlock some genuinely new capabilities.

We created a 1-hour live-coding tutorial to get started in imaging problems with AI, using the deepinverse library youtu.be/YRJRgmXV8_I?...

I’ll be at EurIPS in Copenhagen in early December ! Always up for chats about diffusion, flow matching, Earth observation, AI4climate, etc... Ping me if you’re going! 🇩🇰🌍

I first came across the idea of learning curved interpolants in "Branched Schrödinger Bridge Matching" arxiv.org/abs/2506.09007. I liked it, but I’m curious how well it scales to high-dim settings and how challenging it is to learn sufficiently good interpolants to train the diffusion bridge

"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.

I'm on my way to @caltech.edu for an AI + Science conference. Looking forward to seeing some friends and meeting new ones. There will be a livestream. aiscienceconference.caltech.edu

“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

💥 DeepInverse is now part of the official PyTorch Landscape💥 We are excited to join an ecosystem of great open-source AI libraries, including @hf.co diffusers, MONAI, einops, etc. pytorch.org/blog/deepinv...

🌀🌀🌀New paper on the generation phases of Flow Matching arxiv.org/abs/2510.24830 Are FM & diffusion models nothing else than denoisers at every noise level? In theory yes, *if trained optimally*. But in practice, do all noise level equally matter? with @annegnx.bsky.social, S Martin & R Gribonval

Want to work on generative models and Earth Observation? 🌍 I'm looking for: 🧑‍💻 an intern on generative models for change detection 🧑‍🔬 a PhD student on neurosymbolic generative models for geospatial data Both starting beginning of 2026. Details are below, feel free to email me!

We introduce MIRO: a new paradigm for T2I model alignment integrating reward conditioning into pretraining, eliminating the need for separate fine-tuning/RL stages. This single-stage approach offers unprecedented efficiency and control. - 19x faster convergence ⚡ - 370x less FLOPS than FLUX-dev 📉

“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

New paper, with @rkhashmani.me @marielpettee.bsky.social @garrettmerz.bsky.social Hellen Qu. We introduce a framework for generating realistic, highly multimodal datasets with explicitly calculable mutual information. This is helpful for studying self-supervised learning arxiv.org/abs/2510.21686

"The Principles of Diffusion Models" by Chieh-Hsin Lai, Yang Song, Dongjun Kim, Yuki Mitsufuji, Stefano Ermon. arxiv.org/abs/2510.21890 It might not be the easiest intro to diffusion models, but this monograph is an amazing deep dive into the math behind them and all the nuances

I'm excited to share jaxion, a differentiable Python/JAX library for fuzzy dark matter (axions) + gas + stars, scalable on multiple GPUs ⭐️repo: github.com/JaxionProjec... 📚docs: jaxion.readthedocs.io Feedback + collaborations welcome!

Fisher meets Feynman! 🤝 We use score matching and a trick from quantum field theory to make a product-of-experts family both expressive and efficient for variational inference. To appear as a spotlight @ NeurIPS 2025. #NeurIPS2025 (link below)

that and please share/repost the articles you’re interested in (especially if you’re not the author). If I’m following you, I want to see what you’re reading. We don’t need a fancy algorithm if we can discover great research through the curated posts of the people we follow

Strong afternoon session: Ségolène Martin on how to go from flow matching to denoisers (and hopefully come back?) and Claire Boyer on how learning rate and working in latent spaces affect diffusion models

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/mod...

Excited to share SamudrACE, the first 3D AI ocean–atm–sea-ice #climate emulator! 🚀 Simulates 800 years in 1 day on 1 GPU, ~100× faster than traditional models, straight from your laptop 👩‍💻 Collaboration with @ai2.bsky.social and GFDL, advancing #AIforScience with #DeepLearning. tinyurl.com/Samudrace

I'm already waiting for the next generation of "diffusion transformers features are well-suited for discriminative tasks" but with DiT trained with this representation autoencoders and the loop with be closed

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!

"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!

While working on semidiscrete flow matching this summer (➡️ arxiv.org/abs/2509.25519), I kept looking for a video illustrating that the velocity field solving the Benamou-Brenier OT problem is NOT constant w.r.t. time ⏳... so I did it myself, take a look! ott-jax.readthedocs.io/tutorials/th...

🚨Updated: "How far can we go with ImageNet for Text-to-Image generation?" TL;DR: train a text2image model from scratch on ImageNet only and beat SDXL. Paper, code, data available! Reproducible science FTW! 🧵👇 📜 arxiv.org/abs/2502.21318 💻 github.com/lucasdegeorg... 💽 huggingface.co/arijitghosh/...

Very excited to share our preprint: Self-Speculative Masked Diffusions We speed up sampling of masked diffusion models by ~2x by using speculative sampling and a hybrid non-causal / causal transformer arxiv.org/abs/2510.03929 w/ @vdebortoli.bsky.social, Jiaxin Shi, @arnauddoucet.bsky.social

🚀 After more than a year of work — and many great discussions with curious minds & domain experts — we’re excited to announce the public release of 𝐀𝐩𝐩𝐚, our latent diffusion model for global data assimilation! Check the repo and the complete wiki! github.com/montefiore-s...

"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

#Distinction 🏆| Charlotte Pelletier, lauréate d'une chaire #IUF, développe des méthodes d’intelligence artificielle appliquées aux séries temporelles d’images satellitaires. ➡️ www.ins2i.cnrs.fr/fr/cnrsinfo/... 🤝 @irisa-lab.bsky.social @cnrs-bretagneloire.bsky.social

Reposting because part of me wants to see EBM make a comeback and hopes flow-based training can help it scale.

Our two phenomenal interns, Alireza Mousavi-Hosseini and Stephen Zhang @syz.bsky.social have been cooking some really cool work with Michal Klein and me over the summer. Relying on optimal transport couplings (to pick noise and data pairs) should, in principle, be helpful to guide flow matching 🧵

You can learn (condition+time)-dependent weights for classifier-free guidance using reward functions like the CLIP score arxiv.org/abs/2510.00815. I wonder if, for text-to-image models, the temporal evolution of learned weights reveals information about the sizes of objects described in the caption