Xinyu Wang

StatePlay. StatePlay: State-Aware Game World Models for Mechanics-Consistent Generation

OmniSource. 三轨一源 捕捉新知、机会与创造力🍀 科研 · 创业 · 开发者工具 One source, three tracks surfacing new ideas, opportunities, and builder energy across research, venture, and tools

solaris. The first multiplayer video world model in Minecraft

Kimi-K3. Open Frontier Intelligence

AlayaWorld. Full-stack open-source interactive long-horizon world model.

MemoryWAM. MemoryWAM: Efficient World Action Modeling with Persistent Memory

posterly. Build academic conference posters as a single HTML/CSS file, rendered to print-ready PDF via headless Chromium. A coding-agent skill.

LiveLight. [TOG 2026] LiveLight: This repo is the official implementation of "LiveLight : Real-time video relighting via interactive control"

CineDance.

MuSS. A Large-Scale Dataset and Cinematic Narrative Benchmark for Multi-Shot Subject-to-Video Generation

overleaf-codex. An unofficial Chrome extension that brings Codex conversations, project context, file/line references, and safe LaTeX edits into the Overleaf source editor.

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

Self_Gradient_Forcing. Self Gradient Forcing (SGF) recovers the missing context-gradient path for self-generated causal memory through a bounded two-pass replay, enabling models trained with only a 5-second window to extrapolate to minute-scale videos with stronger identity, layout, and temporal stability.

Matrix-Game-3.5. Matrix-Game 3.5: Enhancing Real-Time Streaming Interactive World Models with Patch Memory

video-download-updates. 视频下载桌面更新的公开发布源｜产品官网，详细介绍工具

SCD. [ICML26] Official codebase for "Speculative Coupled Decoding for Fast Autoregressive Visual Generation"

OPSD-V. On-Policy Self-Distillation for Post-Training Few-Step Autoregressive Video Generators

GameGen-X.

lingbot-vision. Self-supervised learning for spatial perception

WorldFoundry. Unified World Model Inference & Evaluation Infrastructure

Awesome-Human-Motion-Video-Generation. 【Accepted by TPAMI】Human Motion Video Generation: A Survey (https://ieeexplore.ieee.org/document/11106267)

UltraFlash. Ultra Flash: Scaling Real-Time Streaming Video Generation to High Resolutions

lingbot-world-v2. Infinite Worlds with Versatile Interactions

UniTemp. [ECCV 2026] UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillation

DissolveStereo. DissolveStereo: Coarse Depth Injection for Zero-Shot Stereo Video Generation [SIGGRAPH/TOG 2026]

Auto-claude-code-research-in-sleep. ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework, no lock-in — works with Claude Code, Codex, OpenClaw, or any LLM agent.

trip-map-builder. 旅行行程规划技能：规划 → 小红书调研 → 交互式地图页面 | Agent Skill for trip planning with 小红书 research and interactive map generation

lyra. Project Lyra: Open Generative 3D World Models

mira. Code for MIRA: Multiplayer Interactive World Models with Representation Autoencoders

next-forcing. Next Forcing: World Action Modeling with Multi-Chunk Prediction (MCP)

ImagineTime.

starVLA. StarVLA: A Lego-like Codebase for Vision-Language-Action Model Developing

diagdistill. Implementation of <Streaming Autoregressive Video Generation via Diagonal Distillation> in ICLR 2026

opentouch. OPENTOUCH: Bringing Full-Hand Touch to Real-World Interaction

OSWorld-V2. OSWorld 2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks

LiveEdit. [ECCV 2026] LiveEdit: Towards Real-Time Diffusion-Based Streaming Video Editing

do-as-i-do. Official Codebase for "Do as I Do: Dexterous Manipulation Data from Everyday Human Videos"

Track2View. Track2View: 4D-Consistent Camera-Controlled Video Generation via Paired 3D Point Tracks

AdaWorld. [ICML'25] The PyTorch implementation of paper: "AdaWorld: Learning Adaptable World Models with Latent Actions".

ForgeWM. Train a real-time, playable Minecraft Video World Models on 8 GPUs — keyboard & mouse control, fully open and reproducible.

LightEWM. Lightweight embodied world model training & inference repo supporting world models, IDM action models, and autoregressive causal modeling.

codex-claude-bridge. Bidirectional bridge between Claude Code and OpenAI Codex CLI. Built on Claude Code Channels. Two AI agents, one conversation, real-time web UI.

awesome-agentic-world-modeling. Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond

OPUS. Code for ICML 2026 Oral paper "OPUS: Towards Efficient and Principled Data Selection in Large Language Model Pre-training in Every Iteration."

EasyVFX. [SIGGRAPH 2026] EasyVFX: This repo is the official implementation of "EasyVFX: Frequency-Driven Decoupling for Resource-Efficient VFX Generation"

lingbot-world. Advancing Open-source World Models

awesome-vla-wam. A Curated List of Vision-Language-Action (VLA) and World Action Models (WAM) Research and Beyond

planning-with-files. Persistent file-based planning for AI coding agents and long-running tasks. Crash-proof markdown plans, session recovery after /clear and compaction, per-turn re-injection against context rot, deterministic completion gate. Manus-style. Claude Code, Codex, Cursor, Kiro, OpenCode and 60+ agents via the Agent Skills standard.

RoboTwin. [ICML 2026] RoboTwin 2.0 Offical Repo

Humanizer-zh. Humanizer 的汉化版本，Claude Code Skills，旨在消除文本中 AI 生成的痕迹。

Research-Paper-Writing-Skills. Skill package for ML/CV/NLP paper writing, curated and adapted from Prof. Peng Sida's open notes for Codex, Claude Code, and Gemini.

GPT-Image2-Skill. GPT Image 2 prompt gallery, image prompt library, agentic skill, and CLI for OpenAI image generation/editing

FollowYourShape. [ICLR 2026] Follow-Your-Shape: This repo is the official implementation of "Follow-Your-Shape: Shape-Aware Image Editing via Trajectory-Guided Region Control"

Supervisor-Skills. 将博导十年科研经验炼化为可直接调用的 AI 技能。从 Idea 构思到论文投稿，你的 AI 科研副导师。

open-oasis. Inference script for Oasis 500M

MultiWorld. Official Implementation of MultiWorld: Scalable Multi-Agent Multi-View Video World Models

street-fighter-ai. This is an AI agent for Street Fighter II Champion Edition.

spmem. [NeurIPS 2025] Video World Models with Long-term Spatial Memory

unrealzoo-gym. [ICCV 2025 Highlights] Large-scale photo-realistic virtual worlds for embodied AI

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

Video2LoRA. Video2LoRA: Unified Semantic-Controlled Video Generation via Per-Reference-Video LoRA （CVPR 2026 Findings）

Awesome-Inference-Time-Scaling. Paper List of Inference/Test Time Scaling/Computing

Awesome-World-Models.

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

awesome_sora2_prompt. Awesome Sora2 Prompt

GroupEditing. [CVPR 2026] Group Editing: This repo is the official implementation of "Group Editing: Edit Multiple Images in One Go"

X-Dub. Try X-Dub to sync any character in a video with any audio you like | Official repository for "From Inpainting to Editing: Unlocking Robust Mask-Free Visual Dubbing via Generative Bootstrapping"

Awesome-Video-Diffusion. A curated list of recent diffusion models for video generation, editing, and various other applications.

HoloCine. [CVPR 2026 Highlight] Official Implementations for Paper - HoloCine: Holistic Generation of Cinematic Multi-Shot Long Video Narratives

HiAR. [ECCV 2026] sink-free, anti-drift causal AR Video Generation

NovaEdit. [CVPR26] Nova: Video Editing via single/multiple frame references

Helios. Helios: Real Real-Time Long Video Generation Model

tttLRM. [CVPR 2026 Highlight] tttLRM: Test-Time Training for Long Context and Autoregressive 3D Reconstruction

taehv. Tiny AutoEncoder for Hunyuan Video (and other video models)

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

PersonaLive. [CVPR 2026] PersonaLive! : Expressive Portrait Image Animation for Live Streaming

Causal-Forcing. [ICML 2026] Official codebase for "Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation" & Causal Forcing++

awesome-ai-research-writing. Elevate your AI research writing, no more tedious polishing ✨

Shot2Story. A new multi-shot video understanding benchmark Shot2Story with comprehensive video summaries and detailed shot-level captions.

Innovator-VL. Fully Open-source Multimodal Language Models for Science Discovery

CycleGAN. Software that can generate photos from paintings, turn horses into zebras, perform style transfer, and more.

img2img-turbo. One-step image-to-image with Stable Diffusion turbo: sketch2image, day2night, and more

VoxHammer. [3DV 2026 Oral] VoxHammer: Training-Free Precise and Coherent 3D Editing in Native 3D Space

FollowYourClick. [AAAI 2025] Follow-Your-Click: This repo is the official implementation of "Follow-Your-Click: Open-domain Regional Image Animation via Short Prompts"

FollowYourPose. [AAAI 2024] Follow-Your-Pose: This repo is the official implementation of "Follow-Your-Pose : Pose-Guided Text-to-Video Generation using Pose-Free Videos"

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

AlphaGPT. 基于深度强化学习的开源自动因子工厂。

EgoEdit. [CVPR 2026] 👋 Dataset and Benchmark code for EgoEdit

FlashVSR. [CVPR 2026] Towards Real-Time Diffusion-Based Streaming Video Super-Resolution — An efficient one-step diffusion framework for streaming VSR with locality-constrained sparse attention and a tiny conditional decoder.

geneval. GenEval: An object-focused framework for evaluating text-to-image alignment

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

StoryMem. Official code for StoryMem: Multi-shot Long Video Storytelling with Memory

RollingForcing. [ICLR 2026] Official Repo for Rolling Forcing: Autoregressive Long Video Diffusion in Real Time

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

Self-Forcing-Plus. Unofficial extension implementation of Self-Forcing to support I2V && 14B training.

DeT. [ICCV 2025] Decouple and Track: Benchmarking and Improving Video Diffusion Transformers for Motion Transfer

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

Awesome-Visual-Grounding. [TPAMI 2025] Towards Visual Grounding: A Survey

RecStore. RecStore: High-performance parameter storage for large-scale recommendation models, unifying heterogeneous memory as a scalable embedding pool.

DiT-MoE. Scaling Diffusion Transformers with Mixture of Experts

must3r. MUSt3R: Multi-view Network for Stereo 3D Reconstruction

World-Model. [ACM CSUR 2025] Understanding World or Predicting Future? A Comprehensive Survey of World Models

DeepFashion-MultiModal. A large-scale high-quality human dataset with rich multi-modal annotations

MV-VTON. [AAAI 2025] MV-VTON: Multi-View Virtual Try-On with Diffusion Models

RoPECraft. [NIPS2025] RoPECraft: Training-Free Motion Transfer with Trajectory-Guided RoPE Optimization on Diffusion Transformers

ALIKED. ALIKED: A Lighter Keypoint and Descriptor Extraction Network via Deformable Transformation

map-anything. MapAnything: Universal Feed-Forward Metric 3D Reconstruction

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

RALU. [CVPR 2026 Highlight] Training-free Mixed-Resolution Latent Upsampling for Spatially Accelerated Diffusion Transformers

Cache4Diffusion. Aiming to integrate most existing feature caching-based diffusion acceleration schemes into a unified framework.

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

Transformers-Tutorials. This repository contains demos I made with the Transformers library by HuggingFace.

In-Context-LoRA. Official repository of In-Context LoRA for Diffusion Transformers

Sci-Fi. Sci-Fi: Symmetric Constraint for Frame Inbetweening

cosmos-transfer1-diffusion-renderer. Cosmos-Transfer1-DiffusionRenderer: High-quality video de-lighting and re-lighting based on Cosmos video diffusion framework

Awesome-Diffusion-Acceleration-Cache. A curated list of research papers, resources, and advancements on Diffusion Cache and related efficient diffusion model acceleration techniques.

Awesome-DLMs. The official GitHub repo for the survey paper "A Survey on Diffusion Language Models".

Diffusion-LLM-Papers. A Collection of Papers on Diffusion Language Models

ArtiMuse. [CVPR 2026] ArtiMuse: Fine-Grained Image Aesthetics Assessment with Joint Scoring and Expert-Level Understanding（书生 · 妙析多模态美学理解大模型）

comfyui-lumi-batcher. ComfyUI Lumi Batcher is a batch processing extension plugin designed for ComfyUI, aiming to improve workflow debugging efficiency. Traditional debugging methods require adjusting parameters one by one, while this tool significantly enhances work efficiency through batch processing capabilities.

Qwen-Image. Qwen-Image is a powerful image generation foundation model capable of complex text rendering and precise image editing.

sd-scripts. Python

HunyuanWorld-1.0. Generating Immersive, Explorable, and Interactive 3D Worlds from Words or Pixels with Hunyuan3D World Model

Pi3. [ICLR 2026] π^3: Permutation-Equivariant Visual Geometry Learning

YUME. The official code of Yume

Overleaf-Bib-Helper. Enhances Overleaf by allowing article searches and BibTeX retrieval from DBLP and Google Scholar | 通过允许从 DBLP 和 Google Scholar 进行文章搜索和获取 BibTeX 来增强 Overleaf。

Awesome-Controllable-Video-Generation. [ArXiv 2025] A survey about controllable video generation: This repo is the official awesome of "Controllable video generation: A survey"

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

MemOS. Self-evolving memory OS for LLM & AI Agents: ultra-persistent memory, hybrid-retrieval, and cross-task skill reuse, with 35.24% token savings

JarvisArt. [NeurIPS' 2025] JarvisArt: Liberating Human Artistic Creativity via an Intelligent Photo Retouching Agent

RegionDrag. [ECCV2024] RegionDrag: Fast Region-Based Image Editing with Diffusion Models

RoboMaster. [ICLR’26] Learning Video Generation for Robotic Manipulation with Collaborative Trajectory Control

NonogramSolver. 基于 C++ 实现的高性能数织（Nonogram）求解器

Awesome-Unified-Multimodal. 📖 This is a repository for organizing papers, codes, and other resources related to unified multimodal models.

Awesome-World-Model. Collect some World Models for Autonomous Driving (and Robotic, etc.) papers.

LeetCUDA. LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.

architecture-windlx-lab. 山东大学计算机体系结构实验报告：Computer Architecture Lab with WinDLX

CompilerDesignStarter. 山东大学编译原理与技术课程新版实验一~三

Sparse-VideoGen. [ICML2025, NeurIPS2025 Spotlight] Sparse VideoGen 1 & 2: Accelerating Video Diffusion Transformers with Sparse Attention

xDiT. xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism

OptiStore. OptiStore: Sequential-Aware Distributed Object Storage System, from 2025华为软件精英挑战赛, team 241MacTeam

Awesome-Efficient-Inference-for-LRMs. [IEEE T-PAMI] Awesome-Efficient-Inference-for-LRMs is a collection of state-of-the-art, novel, exciting, token-efficient methods for Large Reasoning Models (LRMs). It contains papers, codes, datasets, evaluations, and analyses. https://arxiv.org/pdf/2503.23077

tsxt-assignment-solutions. This collection contains all my solutions to coursework assignments from my undergraduate studies. I pursued my undergraduate degree at Taishan College (Tai Shan Xue Tang in Chinese Pinyin), Shandong University.

video-generation-survey. A reading list of video generation

cv-arxiv-daily. 🎓Automatically Update CV Papers Daily using Github Actions

MMVP. Python

datatrove. Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.

TaylorSeer. [ICCV2025] From Reusing to Forecasting: Accelerating Diffusion Models with TaylorSeers

GLIGEN. Open-Set Grounded Text-to-Image Generation

data-centric-AI. A curated, but incomplete, list of data-centric AI resources.

go-cursor-help. 解决Cursor在免费订阅期间出现以下提示的问题: Your request has been blocked as our system has detected suspicious activity / You've reached your trial request limit. / Too many free trial accounts used on this machine.

Awesome-Image-Editing. A Survey of Image Editing

TCSVT25-ViMAEdit. Python

python_314. 314道python选择题，包括python基础与机器学习、神经网络，Flask编译，开源免费

Awesome-Model-Merging-Methods-Theories-Applications. Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunities. ACM Computing Surveys, 2026.

Fk-Diffusion-Steering. A general framework for inference-time scaling and steering of diffusion models with arbitrary rewards.

Blur. In the "2023 Spring Intel oneAPI Campus Hackathon Competition", the project won the third prize and a special award.

prompt-to-prompt. workable version on diffusers 0.17.1

Awesome-Generation-Acceleration. 📚 Collection of awesome generation acceleration resources.

PixArt-alpha. PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis

Embodied-AI-Guide. [Lumina具身智能社区] 具身智能技术指南 Embodied-AI-Guide

self-llm. 《开源大模型食用指南》针对中国宝宝量身打造的基于Linux环境快速微调（全参数/Lora）、部署国内外开源大模型（LLM）/多模态大模型（MLLM）教程

trl. Train transformer language models with reinforcement learning.

webhook. webhook is a lightweight incoming webhook server to run shell commands

CSLabInfo. 关于2025年CS保研实验室/导师招生广告的汇总。欢迎想要打广告的小伙伴积极PR，资瓷一下互联网精神吼不吼啊？

ArXiv_CV_Papers_Daily. Daily feed of the latest Computer Vision research papers from https://arxiv.org.

daam. Diffusion attentive attribution maps for interpreting Stable Diffusion.

start-learning-opengl. Learn opengl together

instant-ngp. Instant neural graphics primitives: lightning fast NeRF and more

VLMEvalKit. Open-source evaluation toolkit of large multi-modality models (LMMs), support 220+ LMMs, 80+ benchmarks

opencompass. OpenCompass is an LLM evaluation platform, supporting a wide range of models (Llama3, Mistral, InternLM2,GPT-4,LLaMa2, Qwen,GLM, Claude, etc) over 100+ datasets.

papers.cool. Cool Papers - Immersive Paper Discovery

GEM. [CVPR24] Official Implementation of GEM (Grounding Everything Module)

Parser. C++

awesome-diffusion-categorized. collection of diffusion model papers categorized by their subareas

MachineTestCode. 考研保研机试

MagicBrush. [NeurIPS'23] "MagicBrush: A Manually Annotated Dataset for Instruction-Guided Image Editing".

HQ-Edit. [ICLR 2025] HQ-Edit: A High-Quality and High-Coverage Dataset for General Image Editing

prompt-to-prompt. Jupyter Notebook

instruct-pix2pix. Python

Machine-Learning-Labs. Jupyter Notebook

King-of-Pigeon. 计算机保研简历与文书实用模板

flux. Official inference repo for FLUX.1 models

gaussian-splatting. Original reference implementation of "3D Gaussian Splatting for Real-Time Radiance Field Rendering"

classifier-free-diffusion-guidance-Pytorch. a simple unofficial implementation of classifier-free diffusion guidance

fast-DiT. Fast Diffusion Models with Transformers

FIDDLE. FlexIble Data-Driven pipeLinE – a preprocessing pipeline that transforms structured EHR data into feature vectors to be used with ML algorithms. https://doi.org/10.1093/jamia/ocaa139

pytorch_federated_learning. PyTorch Federated Learning (easy to use and extend)

minimal-diffusion. A minimal yet resourceful implementation of diffusion models (along with pretrained models + synthetic images for nine datasets)

EluCD. Elucidating The Design Space of Classifier-Guided Diffusion Generation

my-ielts. 雅思词汇真经、雅思语法、听力 179、阅读 538 同义替换等。Everything during preparing for my IELTS exam.

sdu_handwriting. 这是一个DIY山大信纸并生成手写体的教程, 可以让有需要模拟书写的同学快速优雅地在山大信纸上生成模拟手写体.

sd-webui-cads. Greatly increase the diversity of your generated images in Automatic1111 WebUI through Condition-Annealed Sampling.

Awesome-Change-Captioning. 🔥Collection of resources and papers

DiffRS. Official PyTorch implementation for Diffusion Rejection Sampling (DiffRS) in ICML 2024.

DG. Official repo for Discriminator Guidance.

minority-guidance. Official PyTorch Implementation of "Don't Play Favorites: Minority Guidance for Diffusion Models" (ICLR 2024)

OC_LT. Official code base for "Long-Tailed Diffusion Models With Oriented Calibration" ICLR2024

Multimodal-Sarcasm-Explanation-MuSE. Jupyter Notebook

my-sdu. SDUCS 2020