Ceyuan Yang

UniRL. UniRL is a Framework for Unified Multimodal Model Reinforcement Learning

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

sjh-skills. Shell

claude-code-local-for-vscode. Claude Code VS Code extension patched for Force Local mode — run CLI locally, proxy file ops to remote server via VS Code Remote SSH

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

lingbot-world. Advancing Open-source World Models

lingbot-depth. Masked Depth Modeling for Spatial Perception

MagiAttention. A Distributed Attention Towards Linear Scalability for Ultra-Long Context, Heterogeneous Data Training

Kimi-Linear.

pyCuSFM. Python wrapper for the NVIDIA cuSFM library

seed-oss. Python

VeOmni. VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo

SeedVR. Repo for SeedVR2 (ICLR2026) & SeedVR (CVPR2025 Highlight)

SeedVR2. [ICLR2026] SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training

reverse-engineering-gemma-3n. Reverse Engineering Gemma 3n: Google's New Edge-Optimized Language Model

MMaDA. MMaDA - Open-Sourced Multimodal Large Diffusion Language Models (dLLMs with block diffusion, mixed-CoT, unified RL)

Bagel. Open-source unified multimodal model

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

DualPipe. A bidirectional pipeline parallelism algorithm for computation-communication overlap in DeepSeek V3/R1 training.

Wan2.1. Wan: Open and Advanced Large-Scale Video Generative Models

FlashMLA. FlashMLA: Efficient Multi-head Latent Attention Kernels

open-infra-index. Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation

MoBA. MoBA: Mixture of Block Attention for Long-Context LLMs

Step-Video-T2V. Python

Video-Depth-Anything. [CVPR 2025 Highlight] Video Depth Anything: Consistent Depth Estimation for Super-Long Videos

Vchitect-2.0. Vchitect-2.0: Parallel Transformer for Scaling Up Video Diffusion Models

Sa2VA. Official Repo For Pixel-LLM Codebase: Sa2VA (Arxiv-25), SAMTok (CVPR-26), VRT, SaSaSa2VA (1-st solution for LSVOS)

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.

edicho. [ICCV 2025] Edicho: Consistent Image Editing in the Wild

NuTime. Python

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

MagicQuill. [CVPR'25] Official Implementations for Paper - MagicQuill: An Intelligent Interactive Image Editing System

alphafold3. AlphaFold 3 inference pipeline.

Cosmos-Tokenizer. A suite of image and video neural tokenizers

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

videophy. Video Generation, Physical Commonsense, Semantic Adherence, VideoCon-Physics

DiT-MoE. Scaling Diffusion Transformers with Mixture of Experts

lingua. Meta Lingua: a lean, efficient, and easy-to-hack codebase to research LLMs.

monst3r. Official Implementation of paper "MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion"

torchtitan. A PyTorch native platform for training generative AI models

hash-hop. Long context evaluation for large language models

Qwen3-VL. Qwen3-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.

photographic-dctls. This set of DCTL scripts is about experimenting with the math behind color science, including LogC, ACES AP0, Cineon, and more.

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

DeepSeek-V2. DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model

3dpe. [ECCV 2024] 3DPE: Real-time 3D-aware Portrait Editing from a Single Image

mast3r. Grounding Image Matching in 3D with MASt3R

SOLO. [TMLR] Public code repo for paper "A Single Transformer for Scalable Vision-Language Modeling"

anole. [Extended verision ICLR 2025 Blog Track] Anole: An Open, Autoregressive and Native Multimodal Models for Interleaved Image-Text Generation

Kolors. Kolors Team

SEED-Voken. SEED-Voken: A Series of Powerful Visual Tokenizers

1d-tokenizer. This repo contains the code for 1D tokenizer and generator

aura-sr. AuraSR: GAN-based Super-Resolution for real-world

zeroverse. Official code for NeurIPS 2024 paper LRM-Zero: Training Large Reconstruction Models with Synthesized Data

npq-vit. [ICLR 2025] Binary Spherical Quantization + [CVPR 2026] Leech Spherical Quantization

AsyncDiff. [NeurIPS 2024] AsyncDiff: Parallelizing Diffusion Models by Asynchronous Denoising

LlamaGen. Autoregressive Model Beats Diffusion: 🦙 Llama for Scalable Image Generation

ChronoDepth. ChronoDepth: Learning Temporally Consistent Video Depth from Video Diffusion Priors

Dimba. Transformer-Mamba Diffusion Models

DreamPhysics. [AAAI 2025] DreamPhysics: Learning Physics-Based 3D Dynamics with Video Diffusion Priors

custom-diffusion360. CustomDiffusion360: Customizing Text-to-Image Diffusion with Camera Viewpoint Control

Omost. Your image is almost there!

EasyAnimate. 📺 An End-to-End Solution for High-Resolution and Long Video Generation Based on Transformer Diffusion

edm2. EDM2 and Autoguidance -- Official PyTorch implementation

3DitScene. [ICLR 2025] 3DitScene: Editing Any Scene via Language-guided Disentangled Gaussian Splatting

CameraCtrl. Python

HunyuanDiT. Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding

piecewise-rectified-flow. PeRFlow: Piecewise Rectified Flow as Universal Plug-and-Play Accelerator (NeurIPS 2024)

TeCH. [3DV 2024] Official repo of "TeCH: Text-guided Reconstruction of Lifelike Clothed Humans"

PuLID. [NeurIPS 2024] Official code for PuLID: Pure and Lightning ID Customization via Contrastive Alignment

lightplane. Lightplane implements a highly memory-efficient differentiable radiance field renderer, and a module for unprojecting features from images to 3D grids.

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

PLLaVA. Official repository for the paper PLLaVA

DROID-SLAM. Python

pixel-perfect-sfm. Pixel-Perfect Structure-from-Motion with Featuremetric Refinement (ICCV 2021, Best Student Paper Award)

Hierarchical-Localization. Visual localization made easy with hloc

InternVL. [CVPR 2024 Oral] InternVL Family: A Pioneering Open-Source Alternative to GPT-4o. 接近GPT-4o表现的开源多模态对话模型

vggsfm. VGGSfM: Visual Geometry Grounded Deep Structure From Motion

Lumina-T2X. Lumina-T2X is a unified framework for Text to Any Modality Generation

llama3. The official Meta Llama 3 GitHub site

Portrait-Mode-Video. Video dataset dedicated to portrait-mode video recognition.

mickey. [CVPR 2024 - Oral] Matching 2D Images in 3D: Metric Relative Pose from Metric Correspondences

SpaTracker. [CVPR 2024 Highlight] Official PyTorch implementation of SpatialTracker: Tracking Any 2D Pixels in 3D Space

VGen. Official repo for VGen: a holistic video generation ecosystem for video generation building on diffusion models

Diffusion-RWKV. Scaling RWKV-Like Architectures for Diffusion Models

MiraData. Official repo for paper "MiraData: A Large-Scale Video Dataset with Long Durations and Structured Captions"

ring-attention. ring-attention experiments

GRM. Large Gaussian Reconstruction Model for Efficient 3D Reconstruction and Generation

ect. Consistency Models Made Easy

VidProM. [NeurIPS 2024] VidProM: A Million-scale Real Prompt-Gallery Dataset for Text-to-Video Diffusion Models

FRESCO. [CVPR 2024] FRESCO: Spatial-Temporal Correspondence for Zero-Shot Video Translation

transformer-debugger. Python

Panda-70M. [CVPR 2024] Panda-70M: Captioning 70M Videos with Multiple Cross-Modality Teachers

ringattention. Large Context Attention

TransformerEngine. A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on Hopper, Ada and Blackwell GPUs, to provide better performance with lower memory utilization in both training and inference.

Vespa. Video Diffusion State Space Models

VMamba. VMamba: Visual State Space Models，code is based on mamba

AdaMPI. [SIGGRAPH 2022] Single-View View Synthesis in the Wild with Learned Adaptive Multiplane Images

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

LLaMA2-Accessory. An Open-source Toolkit for LLM Development

dust3r. DUSt3R: Geometric 3D Vision Made Easy

VideoSys. VideoSys: An easy and efficient system for video generation

ring-flash-attention. Ring attention implementation with flash attention

DiS. Scalable Diffusion Models with State Space Backbone

mamba. Mamba SSM architecture

gemma_pytorch. The official PyTorch implementation of Google's Gemma models

attention_with_linear_biases. Code for the ALiBi method for transformer language models (ICLR 2022)

StableCascade. Official Code for Stable Cascade

LaVIT. LaVIT: Empower the Large Language Model to Understand and Generate Visual Content

SinMPI. Pytorch implementation of SinMPI (SIGGRAPH Asia 2023)

ColossalAI. Making large AI models cheaper, faster and more accessible

InternEvo. InternEvo is an open-sourced lightweight training framework aims to support model pre-training without the need for extensive dependencies.

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

Moore-AnimateAnyone. Character Animation (AnimateAnyone, Face Reenactment)

gpt_academic. 为GPT/GLM等LLM大语言模型提供实用化交互接口，特别优化论文阅读/润色/写作体验，模块化设计，支持自定义快捷按钮&函数插件，支持Python和C++等项目剖析&自译解功能，PDF/LaTex论文翻译&总结功能，支持并行问询多种LLM模型，支持chatglm3等本地模型。接入通义千问, deepseekcoder, 讯飞星火, 文心一言, llama2, rwkv, claude2, moss等。

ml-aim. This repository provides the code and model checkpoints for AIMv1 and AIMv2 research projects.

compose-and-conquer. [ICLR 2024] Official repo. for Compose and Conquer: Diffusion-Based 3D Depth Aware Composable Image Synthesis

pycolmap. (Deprecated) Python bindings for COLMAP

alphageometry. Python

LongLoRA. Code and documents of LongLoRA and LongAlpaca (ICLR 2024 Oral)

FreeNoise-AnimateDiff. [ICLR 2024] Code for FreeNoise based on AnimateDiff

FreeInit. [ECCV 2024] FreeInit: Bridging Initialization Gap in Video Diffusion Models

LangSplat. Official implementation of the paper "LangSplat: 3D Language Gaussian Splatting" [CVPR2024 Highlight]

kohya_ss. Python

PoseDiffusion. [ICCV 2023] PoseDiffusion: Solving Pose Estimation via Diffusion-aided Bundle Adjustment

splatter-image. Official implementation of `Splatter Image: Ultra-Fast Single-View 3D Reconstruction' CVPR 2024

OpenLRM. An open-source impl. of Large Reconstruction Models

4dfy. 4D-fy: Text-to-4D Generation Using Hybrid Score Distillation Sampling

Upscale-A-Video. [CVPR 2024] Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution

SceneWiz3D. [CVPR 2024] SceneWiz3D: Towards Text-guided 3D Scene Composition

GenDeF. Python

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

MotionCtrl. Official Code for MotionCtrl [SIGGRAPH 2024]

AnimateZero. Official PyTorch implementation for the paper "AnimateZero: Video Diffusion Models are Zero-Shot Image Animators"

BerfScene. [CVPR 2024] BerfScene: Bev-conditioned Equivariant Radiance Fields for Infinite 3D Scene Generation

magic-animate. [CVPR 2024] Official repository for "MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model"

sd-webui-panorama-viewer. Sends rendered SD_auto1111 images quickly to this panorama (hdri, equirectangular) viewer

VBench. [CVPR2024 Highlight] VBench - We Evaluate Video Generation

SEINE. [ICLR 2024] SEINE: Short-to-Long Video Diffusion Model for Generative Transition and Prediction

LaVie. [IJCV 2024] LaVie: High-Quality Video Generation with Cascaded Latent Diffusion Models

AnimateAnyone. Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation

CKA-Evaluation. Code for our metric paper: Revisiting the Evaluation of Image Synthesis with GANs [NeurIPS D&B '2023]

PhysGaussian. [CVPR 2024 Highlight] PhysGaussian: Physics-Integrated 3D Gaussians for Generative Dynamics

generative-models. Generative Models by Stability AI

latent-consistency-model. Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference

helm. Holistic Evaluation of Language Models (HELM) is an open source Python framework created by the Center for Research on Foundation Models (CRFM) at Stanford for holistic, reproducible and transparent evaluation of foundation models, including large language models (LLMs) and multimodal models.

consistencydecoder. Consistency Distilled Diff VAE

awesome-llm-powered-agent. Awesome things about LLM-powered agents. Papers / Repos / Blogs / ...

LAVIS. LAVIS - A One-stop Library for Language-Vision Intelligence

BLIP. PyTorch code for BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation

awesome-ai-painting. AI绘画资料合集（包含国内外可使用平台、使用教程、参数教程、部署教程、业界新闻等等） Stable diffusion、AnimateDiff、Stable Cascade 、Stable SDXL Turbo

consistency_models. Official repo for consistency models.

Stanford-ORB. Official Repository of the Stanford-ORB dataset

4DGaussians. [CVPR 2024] 4D Gaussian Splatting for Real-Time Dynamic Scene Rendering

ComfyUI-AnimateDiff-Evolved. Improved AnimateDiff for ComfyUI and Advanced Sampling Support

AnimateDiff. Official implementation of AnimateDiff.

3d-cinemagraphy. [CVPR 2023] 3D Cinemagraphy from a Single Image

Aurora. Official implementation of Aurora

objaverse-xl. 🪐 Objaverse-XL is a Universe of 10M+ 3D Objects. Contains API Scripts for Downloading and Processing!

InstaFlow. :zap: InstaFlow! One-Step Stable Diffusion with Rectified Flow (ICLR 2024)

ControledAnimateDiff. Controlnet extension of AnimateDiff.

co-tracker. CoTracker is a model for tracking any point (pixel) on a video.

mtm. Official implementation of MTM

magic-edit. MagicEdit: High-Fidelity Temporally Coherent Video Editing

recurrent-interface-network-pytorch. Implementation of Recurrent Interface Network (RIN), for highly efficient generation of images and video without cascading networks, in Pytorch

HAT. CVPR2023 - Activating More Pixels in Image Super-Resolution Transformer TPAMI - HAT: Hybrid Attention Transformer for Image Restoration

animatediff-cli-prompt-travel. animatediff prompt travel

Dynamic3DGaussians. Python

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

awesome-few-shot-generation. [IJCV-2025] A curated list of awesome few-shot image generation papers

text2room. Text2Room generates textured 3D meshes from a given text prompt using 2D text-to-image models (ICCV2023).

Carver. [NeurIPS'23] An efficient PyTorch-based library for training 3D-aware image synthesis models.

BallGAN. [ICCV23] BallGAN: 3D-aware Image Synthesis with a Spherical Background

SEED. Official implementation of SEED-LLaMA (ICLR 2024).

MobileSAM. This is the official code for MobileSAM project that makes SAM lightweight for mobile applications and beyond!

xformers. Hackable and optimized Transformers building blocks, supporting a composable construction.

ReCo. ReCo: Region-Controlled Text-to-Image Generation, CVPR 2023

infinigen. Infinite Photorealistic Worlds using Procedural Generation

StyleSV. [ICLR 2023] Towards Smooth Video Composition

GITM. Ghost in the Minecraft: Generally Capable Agents for Open-World Environments via Large Language Models with Text-based Knowledge and Memory

DragGAN. Official Code for DragGAN (SIGGRAPH 2023)

learning_research. 本人的科研经验

MOSS. An open-source tool-augmented conversational language model from Fudan University

LipsFormer. Python

genvs.

InternImage. [CVPR 2023 Highlight] InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions

evals. Evals is a framework for evaluating LLMs and LLM systems, and an open-source registry of benchmarks.

LoRA. Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models"

mage. A PyTorch implementation of MAGE: MAsked Generative Encoder to Unify Representation Learning and Image Synthesis

TaskMatrix. Python

pv3d. Python

bob-plugin-openai-translator. 基于 LLM 的文本翻译、文本润色、语法纠错 Bob 插件，让我们一起迎接不需要巴别塔的新时代！Licensed under CC BY-NC-SA 4.0

3dgp. 3D generation on ImageNet [ICLR 2023]

ControlNet. Let us control diffusion models!

GALIP. [CVPR2023] A faster, smaller, and better text-to-image model for large-scale training

offsite-tuning. Offsite-Tuning: Transfer Learning without Full Model

POT. POT : Python Optimal Transport

NeRV. Official Pytorch implementation for video neural representation (NeRV)

FFNeRV. FFNeRV: Flow-Guided Frame-Wise Neural Representations for Videos

stylegan-t. [ICML'23] StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis