NFMs combine few step sampling with tractable likelihoods and we are particularly excited about their application to RL post-training 🤖 and Boltzmann Generators 🧪. Come check it out if you're in Sydney for the STODY workshop!🔥

🚨 New paper alert: Martingale Posterior Cox Processes 📄 Paper: arxiv.org/abs/2610.24787 💻 Code: github.com/joshuarooijakkers… 🧵 1/5 (↓)

You can check our recent paper: nitter.siddz.com/bose_joey/status/21121… It's still toyish (pixel space imagenet-64) but we use pixel space likelihoods

🚨 Do you always have multiple rewards but don't know how to steer towards an expensive true reward? Don't fret, friends; we have the paper for you! 📣 Steered Not Stirred: Effective Reward Tilting using Surrogate Reward Mixing, led by the incredible @alec_sargood28 and team The Martini Algorithm 🍸 From 007: everyone now has a license to tilt 🤵 📜 Paper: arxiv.org/abs/2610.25502 🔗 Blog: alecsargood.github.io/martin…

Very happy to share the first paper of my PhD: Normalizing Flow Maps! (NFM)⚡️ We make likelihood evaluation in flow maps fast and tractable by replacing the DiT backbone with a normalizing flow🔄. A big thanks to @tim_hitge, @mmbronstein, @bose_joey and the rest of the team!

I'd like to thank my cracked co-lead @francescoshox and PI @bose_joey for their massive efforts to make NFM come to fruition.

Strong Stochastic Flow Maps has been accepted to #NeurIPS2026 as an Oral! 🎉

🚨As part of the growing OXtal family for ML on Molecular Crystals We are releasing OXtalign 🔮, a fast, open-source Python package for comparing crystal packings. Developed by the juggernaut of a researcher, ChengHao Liu (last author of OXtal)! It computes RMSD_n, the standard packing-similarity metric from CCDC's COMPACK. OXtalign delivers similar accuracy to COMPACK but SIGNIFICANTLY FASTER!!!! Features - Speed: ⚡️ Up to 200x faster and fully deterministic - License 🟢: Comes with a permissive Apache 2.0 license. - Data formats 📂: Reads experimental CIFs, predicted clusters without a unit cell, PDB/XYZ files, and ASE or pymatgen objects. Have fun, crystal people 🚀 Code: github.com/OXtal/oxtalign

I'm attending #ICML2026, presenting a couple papers: 1. "Autoregressive Boltzmann Generators" (Spotlight) arxiv.org/abs/2606.27361 and 2. Few-Step CoFolding with All-Atom Flow Maps---i.e., the DeCAF paper ☕️ (Spotlight) at the GenBio workshop. arxiv.org/pdf/2606.08375 Happy to meet old friends and make new ones! DMs open :)

Protein–ligand cofolding models are getting incredibly powerful… but do they have to be so slow? 🧬🐢💊 Our new preprint introduces a new flow-map framework called DeCAF for fast few-step cofolding — up to 5× faster while preserving sample quality on the SOTA Pearl model and 20x faster than Boltz 1x. ⚡🧵 📜 Blog: genesis.ml/news/genesis-mode… 🔗arXiv: arxiv.org/abs/2606.08375 Code (coming soon): github.com/genesistherapeuti…

New paper 🎉: Coupling Models for One-Step Discrete Generation What is the right route to one-step discrete generation? Recently, the default answer has been: make discrete data continuous, learn a flow map, and then decode back to discrete samples. This is a natural route, but it is not the only one. We propose a different perspective: instead of forcing discrete generation through a continuous flow map, we learn a coupling between discrete sequences and Gaussian noise. Then we train a one-step generator that directly predicts the sequence from the noise using cross-entropy. At inference time, generation is simple: sample Gaussian noise → decode once → obtain a discrete sample. No autoregressive loop. No iterative denoising trajectory. No learning continuous flow model. The key idea is that, for one-step discrete generation, the data–noise coupling should not be treated as a fixed background choice. It is the central object to learn. Empirically, Coupling Models improve strong one-step baselines across text generation, biological sequence generation, and binary image generation. Paper: arxiv.org/pdf/2605.07193 Code: github.com/pengzhangzhi/Coup… Work done with @bose_joey @Anru_Zhang @AlexanderTong7

Our book on Generative AI and Stochastic Thermodynamics can be pre-ordered with a 20% discount until July 31 2027. (All proceeds from the authors will be donated to the African Institute for Mathematical Sciences).

A really excellent book. A few people independently told me this was one of their favorite books over a decade ago. I bought it, and it became one of the textbooks on my shelf I revisit from time to time to spark the joy of holding ideas to a different light. It brings to life the elegance of information theory. A good day to recognize 10 years from the passing of David McKay.

🚨 New paper alert!!! Introducing DISCO!! A new Diffusion model that does CO-design of Enzymes with Joint Diffusion of Structure and Sequence. This is a pretty big deal, because we also have wet-lab results to back it up! Huge shoutout to the whole team, but especially @jarridrb and @ChengHaoLiu1 who are the real stars!

I like diffusion models

Discrete diffusion — but fast? ⚡️ Test-time inference — but for discrete data? 🧠 Categorical Flow Maps: continuous transport toward the simplex, turning discrete generation into a single-step problem. Built on Variational FM (CatFlow), we obtain (self-)distillation from scratch. Language, molecules, and test-time steering — one framework. Scaling to LLMs and foundation models next. Watch this space 👀

Ever since I was a young boy I have had a dream of efficiently sampling from the posterior distribution