David Wessels
@Dafidofff
Followers
422
Following
459
Media
108
Statuses
14K
PhD candidate w/ @erikjbekkers & @egavves interested in Geometric Deep Learning and Generative Modelling at @AmlabUva
Amsterdam
Joined November 2010
🥠 Excited to introduce our latest work on Equivariant Neural Fields (ENFs)! Grounding conditioning variables in geometry 🚀 Paper: https://t.co/LQONG1iQ2o Github: https://t.co/yKXZBLDEJU Project Page: https://t.co/lNpz11Z9Hi Details below 👇👇
3
51
227
Excited to be presenting my first solo-author *spotlight* at NeurIPS on Thursday -- Flow Equivariant RNNS! Exhibit Hall C,D,E #2612 @ 4:30pm - 7:30pm https://t.co/Qpr3gbxsMv
Why do video models handle motion so poorly? It might be lack of motion equivariance. Very excited to introduce: Flow Equivariant RNNs (FERNNs), the first sequence models to respect symmetries over time. Paper: https://t.co/dkk43PyQe3 Blog: https://t.co/I1gpam1OL8 1/🧵
1
33
208
🌊I am at #NeurIPS2025 ☀️San Diego edition and looking forward to meeting people interested in 🔷 GeometricDL, 🪩 Generative methods, 🧬 AI4Science. Ping me to meet up! 📣 Tomorrow I will be presenting our work on Equivariance and symmetry breaking in CNNs.
1
3
15
We’re heading to #NeurIPS2025 in San Diego with 9 accepted papers🌴 Check out the full list below and come say hi at the posters! 🧵 0 / 9
1
7
15
✨CAMERA READY UPDATE✨ with new cool plots in which we show how we can use our Equivariant Neural Eikonal Solver for path planning in Riemannian manifolds Check our paper here https://t.co/3QnB40p3tE And see you at NeurIPS 🥰
🌍 From earthquake prediction to robot navigation - what connects them? Eikonal equations! We developed E-NES: a neural network that leverages geometric symmetries to solve entire families of velocity fields through group transformations. Grid-free and scalable! 🧵👇
2
24
219
Cool news: our extended Riemannian Gaussian VFM paper is out! 🔮 We define and study a variational objective for probability flows 🌀 on manifolds with closed-form geodesics. @FEijkelboom @a_ppln @CongLiu202212 @wellingmax @jwvdm @erikjbekkers 🔥 📜 https://t.co/PE6I6YcoTn
2
26
53
Today, we are releasing Sonic-3, the fastest, most natural AI voice model out there, together with a guide to clone your own in less than 10m. Give it a try! Details on @krandiash's thread below! 👇
We've raised $100M from Kleiner Perkins, Index Ventures, Lightspeed, and NVIDIA. Today we're introducing Sonic-3 - the state-of-the-art model for realtime conversation. What makes Sonic-3 great: - Breakthrough naturalness - laughter and full emotional range - Lightning fast -
0
2
24
We've raised $100M from Kleiner Perkins, Index Ventures, Lightspeed, and NVIDIA. Today we're introducing Sonic-3 - the state-of-the-art model for realtime conversation. What makes Sonic-3 great: - Breakthrough naturalness - laughter and full emotional range - Lightning fast -
1K
1K
8K
Recent discussions have largely focused on scaling versus geometry. Another perfect example showcasing that geometry could be made scalable, if we as GDL people start to take scaling seriously. Lets take the best of both worlds 🦾🦾
Clifford Algebra Neural Networks are undeservedly dismissed for being too slow, but they don't have to be! 🚀Introducing **flash-clifford**: a hardware-efficient implementation of Clifford Algebra NNs in Triton, featuring the fastest equivariant primitives that scale.
0
1
23
🌶️💎 Thrilled to announce the new paper I collaborated on led by @Mohammad_Niaz94: "Platonic Transformers"! We introduce a novel architecture that extends RoPE-based transformers to symmetry groups. The key breakthrough? It achieves this without suffering any computational
As promised after our great discussion, @chaitanyakjoshi! Your inspiring post led to our formal rejoinder: the Platonic Transformer. What if the "Equivariance vs. Scale" debate is a false premise? Our paper shows you can have both. 📄 Preprint: https://t.co/kd8MFiOmuG 1/9
0
0
15
🌎🦾 A very cool collaboration extending ENFs to solving Eikonal equations! Interested in robotics, earth science or modelling functions over complex manifolds with neural fields, check it out!! 🔥 Congratz Alejandro, great work!!
Now accepted at NeurIPS 🥳
0
1
7
Why do video models handle motion so poorly? It might be lack of motion equivariance. Very excited to introduce: Flow Equivariant RNNs (FERNNs), the first sequence models to respect symmetries over time. Paper: https://t.co/dkk43PyQe3 Blog: https://t.co/I1gpam1OL8 1/🧵
9
80
446
🎉Happy to be in 🇨🇦Vancouver in the summer for ✨ICML2025! Ping me if you want to chat about Symmetries, GDL, Geometric representations + AI4Science, or want to look for the best ramen in town🍜! 🥁Excited to present a few exciting works at the main conference and workshops!
1
6
57
Flow Matching (FM) is one of the hottest ideas in generative AI - and it’s everywhere at #ICML2025. But what is it? And why is it so elegant? 🤔 This thread is an animated, intuitive intro into (Variational) Flow Matching - no dense math required. Let's dive in! 🧵👇
110
258
2K
🔥🚀 Very excited work, extending Equivariant Neural fields to solving velocity fields with a scalable grid-free Eikonal solver! Credits to the amazing @algarciacast for this great work!
🌍 From earthquake prediction to robot navigation - what connects them? Eikonal equations! We developed E-NES: a neural network that leverages geometric symmetries to solve entire families of velocity fields through group transformations. Grid-free and scalable! 🧵👇
1
1
14
Thrilled about the work my internship company @newtheoryai is doing! They're tackling the future of AI beyond just scaling up. Join them as a Senior AI Researcher & help build foundational models grounded in symmetry & geometry for a more efficient, sustainable deep learning era.
Excited to share our first open position at @newtheoryai. Come join the early team as a Senior AI Researcher. We’re building new foundational architectures to understand the world grounded in symmetry, geometry, and mathematical structures ubiquitous in nature. 📍 San Francisco
0
1
6
Excited to be in Singapore next week for #ICLR2025! 🇸🇬 DM if you want to chat about geometric deep learning and/or generative models on manifolds, or just to enjoy some nice specialty coffee ☕️ I don't look like Barbie but my poster does 🪩😎
10
21
128
Einstein was wrong, new theory of relativity E = m c² * ε* φ
Incredible stuff--they wanted to make the thing seem more sophisticated than it is so they threw in two Greek letters but selected values that cancel out so it's still just trade deficit divided by imports.
0
1
10