Learning to Drive via Real-World Simulation at Scale

Visual Geometry Transformer for Autonomous Driving

Official implementation of Don’t Blind Your VLA: Aligning Visual Representations for OOD Generalization. https://blind-vla-paper.github.io

WorldPlay: Interactive World Modeling with Real-Time Latency and Geometric Consistency

Native and Compact Structured Latents for 3D Generation

An implementation of chunked, compressed, N-dimensional arrays for Python.

Official implementation of "RoboTracer: Mastering Spatial Trace with Reasoning in Vision-Language Models for Robotics"

Code repository of "GenieDrive: Towards Physics-Aware Driving World Model with 4D Occupancy Guided Video Generation"

Official code of Motus: A Unified Latent Action World Model

VLA-0: Building State-of-the-Art VLAs with Zero Modification

A unified inference and post-training framework for accelerated video generation.

Any4D: Unified Feed-Forward Metric 4D Reconstruction

Code release for https://wonderzoom.github.io/

Visionary: The World Model Carrier Built on WebGPU-Powered Gaussian Splatting Platform

The official repository of "Astra : General Interactive World Model with Autoregressive Denoising"

[NeurIPS 2025] AutoSeg3D, online real-time 3D segmentation as instance tracking with long-short term query memory for embodied perception

This repo contains the python code as well as the webpage html files for the Lang3D-XL project.

"Paper2Slides: From Paper to Presentation in One Click"

verl: Volcano Engine Reinforcement Learning for LLMs

Official implementation of “4D LangVGGT: 4D Language-Visual Geometry Grounded Transformer”

Light-X: Generative 4D Video Rendering with Camera and Illumination Control

DGGT: Feedforward 4D Reconstruction of Dynamic Driving Scenes using Unposed Images

VLA-Adapter: An Effective Paradigm for Tiny-Scale Vision-Language-Action Model

Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillation

Official implementation of "C3G: Learning Compact 3D Representations with 2K Gaussians"

On-device Image Generation for Apple Silicon

Official implementation of "CAMEO: Correspondence-Attention Alignment for Multi-View Diffusion Models"