Sometimes the simple things in life are just good. ❤️

predictions are hard, especially about the future... 🤷🏼‍♂️

Hiring for BioEmu project @MSFTresearch AI for Science. Berlin DE or Cambridge UK. aka.ms/bioemu-ml - #MachineLearning Researcher aka.ms/bioemu-data - Data Engineer aka.ms/bioemu-applied - Applied Researcher aka.ms/bioemu-princip… - Principal Applied Researcher

Reinforcement Learning (RL) has long been the dominant method for fine-tuning, powering many state-of-the-art LLMs. Methods like PPO and GRPO explore in action space. But can we instead explore directly in parameter space? YES we can. We propose a scalable framework for

"Equivariance matters even more at larger scales" ~ arxiv.org/abs/2510.09768 All the more reason we need scalable architectures with symmetry awareness. I know this is an obvious ask but I'm still confident that scaling and inductive bias need not be at odds. This paper

🤦‍♂️

meine liebste grammatikalische Konstruktion ist das „politische Wir“, das in Sätzen wie „wir müssen wehrfähig werden“ alle meint außer die sprechende Person, aber in Sätze wie „wir müssen unseren Wohlstand bewahren“ nur die sprechende Person, sehr interessant

Super interesting work!

Yet another NequIP in the top 10 with EquFlash, this time with some very clever accelerations! Bringing the total to ….? I’ll leave it to you how to count :) One question this raises is what a lot of folks have told me recently both on here and in private: they find it

We’ve identified a “Collaboration Gap” in today’s top AI models. Testing 32 leading LMs on our novel maze-solving benchmark, we found that models that excel solo can see their performance *collapse* when required to collaborate – even with an identical copy of themselves. A \🧵

😅🌹✊

I decided that Twitter is becoming more and more useless for anything professional/science related. As such I will start to include more shit posting and recreational stuff from now on. Hope it's not taken as an offense by all the serious scientists following me so far :)

👍

(1/n) Can diffusion models simulate molecular dynamics instead of generating independent samples? In our NeurIPS2025 paper, we train energy-based diffusion models that can do both: - Generate independent samples - Learn the underlying potential 𝑼 🧵👇 arxiv.org/abs/2506.17139

Interpretable #MachineLearning reveals the key physical interactions learned by coarse-grained MD models. By Cecilia Clementi and team, led by Klara Bonneau. nature.com/articles/s4146…