Deep models that respect symmetries have many advantages, but might suffer from loss of expressive power. In our recent paper, led by Nadav Dym, we characterize the expressive power of rotation-equivariant point cloud networks. arxiv.org/pdf/2010.02449…

Our paper: "Multiview Neural Surface Reconstruction by Disentangling Geometry and Appearance" has been accepted to NeurIPS (spotlight) ! Check out the code - github.com/lioryariv/idr (Where you can also switch between trained appearances like these examples)

In this #ICML2021 paper we introduce the PHASE loss for training implicit neural representations while incorporating useful inductive bias such as minimal area, unifying popular implicit representations. Paper: arxiv.org/pdf/2106.07689… Talk in #3DGV: youtu.be/lmjpyIWIsZg

Excited to share "Volume Rendering of Neural Implicit Surfaces" (VolSDF): a volume rendering framework for implicit neural surfaces, allowing to learn high fidelity geometry from a sparse set of input images. with Jiatao Gu Yoni Kasten Yaron Lipman arxiv.org/abs/2106.12052 (1/8)

Excited to share that our paper "Moser Flow: Divergence-based Generative Modeling on Manifolds" won an Outstanding Paper Award at NeurIPS 2021!! blog.neurips.cc/2021/11/30/ann… Noam Rozen Aditya Grover Maximilian Nickel See thread for a short summary 👇

*Frame Averaging for Invariant and Equivariant Network Design* #ICLR2022 by Omri Puny Matan Heli Ben-Hamu Ishan Misra Aditya Grover Yaron Lipman Cool paper on designing invariant/equivariant models wrt known symmetries. Let me try a quick breakdown. 👇 arxiv.org/abs/2110.03336

In our new ICML paper, we introduce CNFM: Continuous Normalizing Flow Matching. A new approach for scalable CNFs on manifolds, the first generative model to scale to moderately high dimensional manifolds! arxiv.org/abs/2207.04711 (1/n)

Excited to share "MultiDiffusion"! A controlled image generation framework w/ pre-trained text-to-image diffusion model. * Spatial guidance controls (bounding boxes/masks) * Arbitrary aspect ratios (huge Panoramas!) NO training NO finetuning. [1/3]Lior Yariv Yaron Lipman Tali Dekel

Thrilled to share Mosaic-SDF (M-SDF), a simple 3D representation suitable for 3D generative models! Check out more results of text-to-3D generations here - lioryariv.github.io/msdf

Thanks for sharing our work! joint work with Omri Puny, Itai Gat, Brian Karrer, Uriel Singer and Yaron Lipman

Presenting M-SDF this week at #CVPR2024 ! Come to our poster on Wednesday morning to say hi and learn/hear about training Flow Matching generative model with our new 3D representation

I'll be presenting D-Flow in the upcoming poster session at 11:30. Come say hi at poster #500! #ICML2024

A nice blog on the connections between diffusion and Gaussian Flow Matching! If you're interested in what the Flow Matching perspective (including SI and RecFlows) offers beyond the Gaussian case, join our Flow Matching Tutorial at NeurIPS Conference next week.

A new (and comprehensive) Flow Matching guide and codebase released! Join us tomorrow at 9:30AM NeurIPS Conference for the FM tutorial to hear more... arxiv.org/abs/2412.06264 github.com/facebookresear…

Our **Flow Matching Tutorial** from #NeurIPS2024 is now publicly available: neurips.cc/virtual/2024/t… Heli Ben-Hamu Ricky T. Q. Chen

Nvidia just dropped FFN Fusion Rethinking Sequential Computation in Large Language Models

[1/n] New paper alert! 🚀 Excited to introduce 𝐓𝐫𝐚𝐧𝐬𝐢𝐭𝐢𝐨𝐧 𝐌𝐚𝐭𝐜𝐡𝐢𝐧𝐠 (𝐓𝐌)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative model🤯, achieving SOTA text-2-image generation! Uriel Singer Itai Gat Yaron Lipman

**Transition Matching** is a new iterative generative paradigm using Flow Matching or AR models to transition between generation intermediate states, leading to an improved generation quality and speed!