ImitSAT: Boolean Satisfiability via Imitation Learning.

Python library for the differentiable hypergeometric distribution

EfficientFlow: Efficient Equivariant Flow Policy Learning for Embodied AI

[ICLR 2026] Code of "MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation"

[CoRL 2025] Repository relating to "TrackVLA: Embodied Visual Tracking in the Wild"

Code for the Molmo Vision-Language Model

JAX implementation of MeanFlow

🤗 LeRobot: Making AI for Robotics more accessible with end-to-end learning

Optimization on Manifolds Spring 2023 Project 2: Estimating rotations from relative measurements

TetSphere Splatting: Representing High-Quality Geometry with Lagrangian Volumetric Meshes

[CVPR 2025 Best Paper Award] VGGT: Visual Geometry Grounded Transformer

[CVPR 2025] Official repository for “MagicArticulate: Make Your 3D Models Articulation-Ready”

Notes for the Numerics of Machine Learning Lecture Course at the University of Tübingen

[ECCV2024 - Oral, Best Paper Award Candidate] SEA-RAFT: Simple, Efficient, Accurate RAFT for Optical Flow

[CVPR2025] SPARS3R: Semantic Prior Alignment and Regularization for Sparse 3D Reconstruction

Awesome Lists for Tenure-Track Assistant Professors and PhD students. (助理教授/博士生生存指南)

Enjoy the magic of Diffusion models!

[CVPR 2024 - Oral, Best Paper Award Candidate] Marigold: Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

Code for the paper "Low Latency Automotive Vision with Event Cameras", published in Nature

LLM training in simple, raw C/CUDA

"SMPLer: Taming Transformers for Monocular 3D Human Shape and Pose Estimation", TPAMI 2024

[CVPR 2024] Code for SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes

Ofiicial GoodDrag implementation.

[ICML 2024] 3D-VLA: A 3D Vision-Language-Action Generative World Model

Computes the 3D Delaunay triangulation on the GPU

InstantID: Zero-shot Identity-Preserving Generation in Seconds 🔥

Dense Optical Tracking: Connecting the Dots