Jiahang Wang

TripoSR. TripoSR: Fast 3D Object Reconstruction from a Single Image

Kimi-K3. Open Frontier Intelligence

autoresearch. AI agents running research on single-GPU nanochat training automatically

Step-3.5-Flash. Fast, Sharp & Reliable Agentic Intelligence

tau-0-vla. This repo is the official implementation of "τ0-VLA: a Hierarchical Robot Foundation Model with World-Model-Guided Test-Time Computation".

RPent. RPent: Agentic Infrastructure for the Physical World

WorldFoundry. Unified World Model Inference & Evaluation Infrastructure

physical-ai-bench. [CVPR 2026 Oral] PAI-Bench: A Comprehensive Benchmark for Physical AI

NATTEN. Fast Multi-dimensional Sparse Attention

SLA. SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse–Linear Attention

ReflectVLN. ReflectVLN: Training Vision-Language Navigation Agents with Self-Reflective Reasoning

ultralytics. Ultralytics YOLO26, YOLO11, YOLOv8 — object detection, instance segmentation, semantic segmentation, image classification, pose estimation, object tracking

Ctrl-World. ICLR 2026 Paper: Ctrl-World

ABot-World. Infinite Interactive World Rollout on a Single Desktop GPU

flashdreams. high-performance inference and serving library for interactive autoregressive video and world models

lingbot-video. Scaling Mixture-of-Experts Video Pretraining for Embodied Intelligence

NeoVerse. [CVPR 2026 Highlight & Best Paper of VideoWorldModel Workshop] NeoVerse: Enhancing 4D World Model with in-the-wild Monocular Videos

AgiBotWorldChallengeICRA2026-WorldModelBaseline. Baseline Model Repo for World Model Track in AgiBot World Challenge@ICRA 2026

T2I-Distill. [Tutorial] Few-Step Distillation for Text-to-Image Generation: A Practical Guide

RynnWorld-4D. RynnWorld-4D: 4D Embodied World Models for Robotic Manipulation

RynnBrain. RynnBrain 1.1: Towards More Capable and Generalizable Embodied Foundation Models

giga-world-1. A Roadmap to Build World Models for Robot Policy Evaluation

lingbot-vision. Self-supervised learning for spatial perception

lingbot-vla-v2. From Foundation to Application

AffordVLA. Afford-VLA: Action-Aligned Visual Planning via Internalized Affordance

PhysisForcing. PhysisForcing: Physics Reinforced World Simulator for Robotic Manipulation

VLAC. VLAC: A Vision-Language-Action-Critic Model for Robotic Real-World Reinforcement Learning

giga-world-policy. GigaWorld-Policy: An Efficient Action-Centered World–Action Model

giga-world-0. GigaWorld-0: World Models as Data Engine to Empower Embodied AI

Infinite-Forcing. Infinite-Forcing: Towards Infinite-Long Video Generation

cosmos-predict2.5. Cosmos-Predict2.5, the latest version of the Cosmos World Foundation Models (WFMs) family, specialized for simulating and predicting the future state of the world in the form of video.

SpargeAttn. [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.

One-Forcing. Official codebase for "One-Forcing: Towards Stable One-Step Autoregressive Video Generation"

4dgen. [ICLR 2026] Codebase for paper "Geometry-aware 4D Video Generation for Robot Manipulation"

unified_video_action. Official PyTorch Implementation of Unified Video Action Model (RSS 2025)

MotionStream. MotionStream: Real-Time Video Generation with Interactive Motion Controls

tau-0-wm. Python

Awesome-Interactive-World-Model. A Comprehensive Survey of Interactive Video World Models

GLM-V. GLM-4.6V/4.5V/4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning

GLM-5. GLM-5: From Vibe Coding to Agentic Engineering

DreamX-World. DreamX-World: A General-Purpose Interactive World Model

Qwen-VLA. The official repository of Qwen-VLA

arx5-sdk. C++ and Python controller for ARX5 robot arm developed by @yihuai-gao (@real-stanford)

ACE-Step-1.5. The most powerful local music generation model that outperforms almost all commercial alternatives, supporting Mac, AMD, Intel, and CUDA devices.

Light-WAM. Codebase for Light-WAM: Efficient World Action Models with State-Fusion Action Decoding.

audio-intelligence. Elucidated Text-To-Audio (ETTA) is a SOTA text-to-audio model with a holistic understanding of the design space and trained with synthetic captions.

ABot-Earth-0.5. Generative 3D Earth Model by AMap-cvlab

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

cosmos-framework. Our inference and training framework to run on the Cosmos Models

Open-d4rt. Python

ReactiveGWM. [Arxiv 2026] ReactiveGWM: Steering NPC in Reactive Game World Models

UltraVideo. [[NeurIPS 2025] UltraVideo: High-Quality UHD Video Dataset with Comprehensive Captions

HumanNet. HumanNet: Scaling Human-centric Video Learning to One Million Hours

ReVidgen. [ICML 2026🔥]Rethinking Video Generation Model for the Embodied World

cosmos. NVIDIA Cosmos is an open platform of world models, datasets, and tools that enables developers to build Physical AI for robots, autonomous vehicles, smart infrastructure, and more.

ESI-Bench. Python

UniVTAC. Jupyter Notebook

MobileAgent. Mobile-Agent: The Powerful GUI Agent Family

sd3.5. Python

Awesome-Interactive-World-Model. Interactive World Model papers organized by core research challenges.

Protenix. Toward High-Accuracy Open-Source Biomolecular Structure Prediction.

ABot-OCR. High-precision document OCR with structured Markdown output

GE-Sim-V2. Genie Envisioner World Simulator 2.0 (GE-Sim 2.0)

Warp-as-History. Warp-as-History: Generalizable Camera-Controlled Video Generation from One Training Video

Dexbotic-RoboChallengeInference. Python

CogVideo. text and image to video generation: CogVideoX (2024) and CogVideo (ICLR 2023)

triattention. TriAttention — Efficient long reasoning with trigonometric KV cache compression. Enables OpenClaw local deployment on memory-constrained GPUs.

Pyramid-Forcing. Official codebase for "Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generation"

Robo3R. Robo3R: Enhancing Robotic Manipulation with Accurate Feed-Forward 3D Reconstruction

maniparena-repo. Python

awesome-reliable-robotics. Robotics research demonstrating reliability and robustness in the real world (continuously updated)

TRELLIS.2. Native and Compact Structured Latents for 3D Generation

TRELLIS. Official repo for paper "Structured 3D Latents for Scalable and Versatile 3D Generation" (CVPR'25 Spotlight).

OpenCLAY. CLAY: A Controllable Large-scale Generative Model for Creating High-quality 3D Assets

clay. High performance UI layout library in C.

MVDream. Multi-view Diffusion for 3D Generation

DCW. [CVPR 2026] Elucidating the SNR-t Bias of Diffusion Probabilistic Models

ABot-Claw. Python

HiDream-O1-Image. Python

droid. Distributed Robot Interaction Dataset.

Open-Sora-Plan. This project aim to reproduce Sora (Open AI T2V model), we wish the open source community contribute to this project.

lingbot-va. [RSS 2026] Causal video-action world model for generalist robot control

DreamDojo. Official Codebase for "DreamDojo: A Generalist Robot World Model from Large-Scale Human Videos" (ICML 2026)

Forcing-KV. Implementation for paper "Forcing-KV: Hybrid KV Cache Compression for Efficient Autoregressive Video Diffusion Models".

minWM. A Minimal and Elegant Framework & Tutorial for Real-Time Interactive World Models

StereoNav. Jupyter Notebook

Being-H. Being-H is BeingBeyond's family of human-centric embodied foundation models.

AR-Diffusion. Python

vggsfm. VGGSfM: Visual Geometry Grounded Deep Structure From Motion

HunyuanVideo. HunyuanVideo: A Systematic Framework For Large Video Generation Model

CameraCtrl. Python

LongCat-Video. Python

WorldForge. [CVPR 2026 Highlight🎉] Official implementation of WorldForge

SpaceTimePilot. [CVPR 2026] SpaceTimePilot: Generative Rendering of Dynamic Scenes Across Space and Time

WorldArena. WorldArena: A Unified Benchmark for Evaluating Perception and Functional Utility of Embodied World Models

LDA-1B. [RSS 2026] LDA-1B: Scaling Latent Dynamics Action Model via Universal Embodied Data Ingestion

UniVideo. [ICLR 2026] UniVideo: Unified Understanding, Generation, and Editing for Videos

NaVILA. [RSS'25] This repository is the implementation of "NaVILA: Legged Robot Vision-Language-Action Model for Navigation"

legged-loco. Low-level locomotion policy training in Isaac Lab

NaVILA-Bench. Vision-Language Navigation Benchmark in Isaac Lab

cosmos-predict2. Cosmos-Predict2 is a collection of general-purpose world foundation models for Physical AI that can be fine-tuned into customized world models for downstream applications.

lyra. Project Lyra: Open Generative 3D World Models

GC-VLN. [CoRL 2025] GC-VLN: Instruction as Graph Constraints for Training-free Vision-and-Language Navigation

WorldMem. [NeurIPS 2025] WorldMem: Long-term Consistent World Simulation with Memory

Archon. The first open-source harness builder for AI coding. Make AI coding deterministic and repeatable.

FrontierNet. [RA-L 2025] FrontierNet: Learning Visual Cues to Explore

DyWA. [ICCV 2025] DyWA:Dynamics-adaptive World Action Model for Generalizable Non-prehensile Manipulation

Stable-Video-Infinity. [ICLR 26 Oral] Stable Video Infinity: Infinite-Length Video Generation with Error Recycling

alpamayo1.5. NVIDIA Alpamayo 1.5 Nano is an open 10B reasoning VLA model for autonomous vehicles with reinforcement-learning enhanced reasoning, navigation guidance, and visual question answering.

UniWorld. UniWorld: High-Resolution Semantic Encoders for Unified Visual Understanding and Generation

Nav-R2. Official Implementation of paper: [Nav-R2:Dual‑Relation Reasoning for Generalizable Open‑Vocabulary Object‑Goal Navigation]

SAPIEN. SAPIEN Embodied AI Platform

VLA-JEPA. [ECCV 2026] VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model

InternVLA-M1. InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot Policy

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

rlds. Jupyter Notebook

robocasa-gr1-tabletop-tasks. Simulation benchmarks of GR1 Tabletop Tasks for GR00T N1

mujoco. Multi-Joint dynamics with Contact. A general purpose physics simulator.

PGSR. [TVCG2024] PGSR: Planar-based Gaussian Splatting for Efficient and High-Fidelity Surface Reconstruction

AgentVLN.

XAgent. An Autonomous LLM Agent for Complex Task Solving

hermes-agent. The agent that grows with you

kimi-cli. Kimi Code CLI is your next CLI agent.

VEGA-3D. [ECCV 2026] Generation Models Know Space: Unleashing Implicit 3D Priors for Scene Understanding

g3D-LF. Official implementation of "g3D-LF: Generalizable 3D-Language Feature Fields for Embodied Tasks" (CVPR'25).

lerobot-annotate. Lightweight web UI for annotating LeRobot datasets

ccf-deadlines. ⏰ Agenticly track worldwide conference deadlines (Website, Python Cli, Wechat Applet)

tab-out. Keep tabs on your tabs. Turn your "New tabs" page into a mission control, so you can close them easily. Built for people who open too many tabs and never close them.

Fast-SAM3D. [ICML 2026] Official Repo for Fast-SAM3D: 3Dfy Anything in Images but Faster

Qwen3.6. Qwen3.6 is the large language model series developed by Qwen team, Alibaba Group.

HY-World-2.0. HY-World 2.0: A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds

ACoT-VLA. [CVPR 2026] Official implementation of "ACoT-VLA: Action Chain-of-Thought for Vision-Language-Action Models"

video-prediction-policy. Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations https://video-prediction-policy.github.io

cosmos-policy. Cosmos Policy

Reward-Forcing. [CVPR 2026 Highlight] Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillation

AnyDepth. AnyDepth: Depth Estimation Made Easy

q8_kernels. Cuda

Awesome-Embodied-World-Model. Awesome paper list and repos of the paper "A comprehensive survey of embodied world models".

InfiniteVGGT. The official implementation of InfiniteVGGT

GeometryForcing. [ICLR26] Official implementation of Geometry Forcing: Marrying Video Diffusion and 3D Representation for Consistent World Modeling

AgiBot-World. [IROS 2025 Best Paper Award Finalist & IEEE TRO 2026] The Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems

Vid2World. Official repository for "Vid2World: Crafting Video Diffusion Models to Interactive World Models" (ICLR 2026), https://arxiv.org/abs/2505.14357

dreamzero. Code to pretrain, fine-tune, and evaluate DreamZero and run sim & real-world evals

Pyramid-Flow. [ICLR 2025] Pyramidal Flow Matching for Efficient Video Generative Modeling

InfinityStar. [NeurIPS 2025 Oral]Infinity⭐️: Unified Spacetime AutoRegressive Modeling for Visual Generation

pillarnext. PillarNeXt: Rethinking Network Designs for 3D Object Detection in LiDAR Point Clouds (CVPR 2023)

epipolar-dpo. Official repo for: Epipolar Geometry Improves Video Generation Models

LTX-2. Official Python inference and LoRA trainer package for the LTX-2 audio–video generative model.

LTX-Video. Official repository for LTX-Video

MAGI-1. MAGI-1: Autoregressive Video Generation at Scale

NOVA. [ICLR 2025] Autoregressive Video Generation without Vector Quantization

SkyReels-V3. SkyReels V3: Multimodal Video Generation Model

cosmos-rl. Cosmos-RL is a flexible and scalable Reinforcement Learning framework specialized for Physical AI applications.

Sana. SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

causalfusion. Python

nunchaku. [ICLR2025 Spotlight] SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models

seraena. WIP Pytorch code for stably training single-step, mode-dropping, deterministic autoencoders

awesome-diffusion-model-in-rl. A curated list of Diffusion Model in RL resources (continually updated)

DreamVLA. [NeurIPS 2025] DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge

HPSv2. Human Preference Score v2: A Solid Benchmark for Evaluating Human Preferences of Text-to-Image Synthesis

HPSv3. Official implementation of HPSv3: Towards Wide-Spectrum Human Preference Score (ICCV2025)

HWM_PLDM. Python

spirit-v1.5. Spirit-v1.5: A Robotic Foundation Model by Spirit AI

ABot-PhysWorld. Python

Infinite-World. [ICML 2026] | Scaling Interactive World Models to 1000-Frame Horizons via Pose-Free Hierarchical Memory

TransArch. Design hardware-friendly model architectures and migrate existing LLMs with minimal performance loss

Open-Sora. Open-Sora: Democratizing Efficient Video Production for All

Hunyuan-0.5B. Python

DiffusionDPO. Code for "Diffusion Model Alignment Using Direct Preference Optimization"

VideoGPA. [ICML'26] VideoGPA is a self-supervised framework that enhances 3D consistency in Video Diffusion Models.

VACE. [ICCV 2025] Official implementations for paper: VACE: All-in-One Video Creation and Editing

sam-3d-objects. SAM 3D Objects

ao. PyTorch native quantization and sparsity for training and inference

cutlass. CUDA Templates and Python DSLs for High-Performance Linear Algebra

AutoFP8. Python

RaBitQ. The repo has been moved to https://github.com/VectorDB-NTU/RaBitQ-Library. [SIGMOD 2024] RaBitQ: Quantizing High-Dimensional Vectors with a Theoretical Error Bound for Approximate Nearest Neighbor Search

TurboDiffusion. TurboDiffusion: 100–200× Acceleration for Video Diffusion Models

TeaCache. Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model

gemm-fp8. High Performance FP8 GEMM Kernels for SM89 and later GPUs.

SageAttention. [ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

youclaw. 🦞 Your AI Personal Assistant. An intelligent assistant with memory, skills, and scheduled tasks that truly understands you . Helping you tackle everything in work and life.

GenRL. Reinforcement Learning Framework for Visual Generation

Turbo-VAED. [AAAI 2026] Turbo-VAED: Fast and Stable Transfer of Video-VAEs to Mobile Devices

DeepForcing. [ICML 2026] Official implementation of "Deep Forcing: Training-Free Long Video Generation with Deep Sink and Participative Compression"

PackForcing.

turboquant. TurboQuant: Near-optimal KV cache quantization for LLM inference (3-bit keys, 2-bit values) with Triton kernels + vLLM integration

VideoX-Fun. 📹 A more flexible framework that can generate videos at any resolution and creates videos from images.

AngelSlim. Model compression toolkit engineered for enhanced usability, comprehensiveness, and efficiency.

LightCompress. [EMNLP 2024 & AAAI 2026] A powerful toolkit for compressing large models including LLMs, VLMs, and video generative models.

ComfyUI-GGUF. GGUF Quantization support for native ComfyUI models

scannetpp. [ICCV 2023 Oral] ScanNet++: A High-Fidelity Dataset of 3D Indoor Scenes

CompACT. Official implementation of "Planning in 8 Tokens: A Compact Discrete Tokenizer for Latent World Model" (CVPR 2026)

DeepGEMM. DeepGEMM: clean and efficient BLAS kernel library on GPU

Q-DiT. [CVPR 2025] Q-DiT: Accurate Post-Training Quantization for Diffusion Transformers

RL3DEdit. [ECCV 2026] RL3DEdit

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

WorldStereo. [CVPR 2026] WorldStereo: Bridging Camera-Guided Video Generation and Scene Reconstruction via 3D Geometric Memories (WorldExpand of HY-World 2.0)

WildWorld. [ECCV 2026] WildWorld: A Large-Scale Dataset for Dynamic World Modeling with Actions and Explicit State toward Generative ARPG

URSA. [ICLR 2026] 🐻 Uniform Discrete Diffusion with Metric Path for Video Generation

Waver. Industry-level video foundation model for unified Text-to-Video (T2V) and Image-to-Video (I2V) generation.