PyTorch Lightning Optical Flow models, scripts, and pretrained weights.

[ICLR2026 - Oral] WAFT: Warping-Alone Field Transforms for Optical Flow

Official codebase for Fast-WAM: Do World Action Models Need Test-time Future Imagination?

RLinf: Reinforcement Learning Infrastructure for Embodied and Agentic AI

One framework to evaluate any VLA model on any robot simulation benchmark.

RoboCasa: Large-Scale Simulation of Everyday Tasks for Generalist Robots

[ICLR 2026 Oral] Latent Particle World Models official repository

[ICLR 2026] RoboInter: A Holistic Intermediate Representation Suite Towards Robotic Manipulation

Causal video-action world model for generalist robot control

A Pragmatic VLA Foundation Model

VAE modified from Descript Audio Codec, which replaces the RVQ with VAE

DACVAE

REALM: A Real-to-Sim Validated Benchmark for Generalization in Robotic Manipulation

[ICLR 2025] SOTA discrete acoustic codec models with 40/75 tokens per second for audio language modeling

Official inference repo for FLUX.1 models

A optimized PyTorch framework for behavior cloning with flow related generative models.

Team Comet's 2025 BEHAVIOR Challenge Codebase

Distribution Matching Variational AutoEncoder (DMVAE)

DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space

BEHAVIOR-1K: a platform for accelerating Embodied AI research. Join our Discord for support: https://discord.gg/bccR5vGFEx

Multimodal Mixture-of-Experts VAE

Official Implementations for Paper - MagicQuillV2: Precise and Interactive Image Editing with Layered Visual Cues

[CVPR'2026] "MM-ACT: Learn from Multimodal Parallel Generation to Act"

Qwen3-omni is a natively end-to-end, omni-modal LLM developed by the Qwen team at Alibaba Cloud, capable of understanding text, audio, images, and video, as well as generating speech in real time.

Official Implementation of "MMaDA-Parallel: Multimodal Large Diffusion Language Models for Thinking-Aware Editing and Generation"

Discrete Diffusion VLA: Bringing Discrete Diffusion to Action Decoding in Vision-Language-Action Policies

EO: Open-source Unified Embodied Foundation Model Series

SAM 3D Objects