SensenGao

VJA. [ICML 26 ORAL] When the Prompt Becomes Visual: Vision-Centric Jailbreak Attacks for Large Image Editing Models

UniWorld-View. Official implementation of UniWorld-View

diffusion-forcing. code for "Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion"

MeanFlowNFT. [arXiv 2026] This is the official PyTorch implementation of "MeanFlowNFT: Bringing Forward-Process RL to Average-Velocity Generators".

DiffusionNFT. [ICLR 2026 Oral] DiffusionNFT: Online Diffusion Reinforcement with Forward Process

LHTB. Long Horizon Terminal Benchmark with Dense Reward Grading

ArtiFixer. Python

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

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

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

Vidu-S1. Vidu S1: A Real-Time Interactive Video Generation Model

RoboDojo. RoboDojo Official Repo

PixWorld. PixWorld: Unifying 3D Scene Generation and Reconstruction in Pixel Space

RDM. Python

GLD. Official implementation of "Repurposing Geometric Foundation Models for Multi-view Diffusion"

PhysisForcing. PhysisForcing: Physics Reinforced World Simulator for Robotic Manipulation

AlignedNorm. [ICML'2026] Official repository of paper titled "AlignedNorm: Prompting Vision–Language Models via Coupled Prompt Field"

RATs. Implementation of paper "Playful Agentic Robot Learning"

DrivingGen. [ICLR 2026] DrivingGen: A Comprehensive Benchmark for Generative Video World Models in Autonomous Driving

tiptop. Official Repo for TiPToP: A Modular Open-Vocabulary Planning System for Robotic Manipulation

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

Echo-Infinity. Official repo for paper "Echo-Infinity: Learnable Evolving Memory for Real-Time Infinite Video Generation"

JoyAI-Echo. JoyAI-Echo: Pushing the Frontier of Long Audio-Visual Generation

REST3D. From a single casual image to a visually consistent and physically stable interactive 3D scene.

Dataset3D. The first "ImageNet" 3D dataset.

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

AgentDoG. A Diagnostic Guardrail Framework for AI Agent Safety and Security

SceneGen. [3DV 2026] "SceneGen: Single-Image 3D Scene Generation in One Feedforward Pass"

PartCrafter. [NeurIPS 2025] PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers

articraft. An Agentic System for Scalable Articulated 3D Asset Generation

PartFlow. PartFlow: two-stage image-conditioned 3D editing (inference code)

T2I-L2P. Code for "L2P: Unlocking Latent Potential for Pixel Generation"

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

PiD. PiD: Fast and High-Resolution Latent Decoding with Pixel Diffusion

humanize. From Automated Idea Factory to Realization

AI-Infra-Auto-Driven-SKILLS. Python

WoG. [ICML 2026] 🏂 World Guidance: World Modeling in Condition Space for Action Generation

semantic-wm. repository for training action-conditioned latent diffusion world models for robot video generation

HiDream-O1-Image. Python

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

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

AnyFlow. Flow Map OPD for AnyStep Video Diffusion

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

FD-Loss. Python

Flow-OPD. Official Repo of "Flow-OPD: On-Policy Distillation for Flow Matching Models"

OPSD. Python

T2PO. 【ICML2026 Spotlight】 T2PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning

SenseNova-U1. SenseNova-U series: Native Unified Paradigm with NEO-unify from the First Principles

tuna-2. Official implementation of Tuna-2: Pixel Embeddings Beat Vision Encoders for Unified Understanding and Generation

World-R1. [ICML 2026] World-R1: Reinforcing 3D Constraints for Text-to-Video Generation

hermes-agent. The agent that grows with you

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

DiT360. [CVPR 2026] Official implementation of "DiT360: High-Fidelity Panoramic Image Generation via Hybrid Training".

lyra. Project Lyra: Open Generative 3D World Models

Fast-dLLM. Fork of "Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding"

ddtree. Python

Awesome-Feed-Forward-3D. An curated list for feed-forward 3D scene modeling, including research directions, datasets, and applications.

DMax. DMax: Aggressive Parallel Decoding for dLLMs

Uni3C. Uni3C: Unifying Precisely 3D-Enhanced Camera and Human Motion Controls for Video Generation [Siggraph Asian 2025]

ThinkSafe. Python

lagernvs. Official code for "LagerNVS Latent Geometry for Fully Neural Real-time Novel View Synthesis" (CVPR 2026)

worldmesh. [ECCV 2026] WorldMesh: Generating Navigable Multi-Room 3D Scenes via Mesh-Conditioned Image Diffusion

video-subtitle-remover. 基于AI的图片/视频硬字幕去除、文本水印去除，无损分辨率生成去字幕、去水印后的图片/视频文件。无需申请第三方API，本地实现。AI-based tool for removing hard-coded subtitles and text-like watermarks from videos or Pictures.

Multimodal-RAG-Survey-For-Document. [ACL2026 Main] Scaling Beyond Context: A Survey of Multimodal Retrieval-Augmented Generation for Document Understanding

sglang. SGLang is a high-performance serving framework for large language models and multimodal models.

ETA. [ICLR 2025] PyTorch Implementation of "ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time"

Omni-WorldBench. A comprehensive benchmark specifically designed to evaluate the interactive response capabilities of world models in 4D settings.

OneWorld. OneWorld: Taming Scene Generation with 3D Unified Representation Autoencoder

OpenWorldLib. Unified Codebase for Advanced World Models.

PixelGen. Official repository for “PixelGen: Improving Pixel Diffusion with Perceptual Loss”

inspatio-world. Python

SkillJect. SkillJect: Automating Stealthy Skill-Based Prompt Injection for Coding Agents with Trace-Driven Closed-Loop Refinement

sglang. SGLang is a high-performance serving framework for large language models and multimodal models.

SPRVLA. Python

worldfm. Python

RobustVLA. Python

open-lvsm. Python

TransferAttack. TransferAttack is a pytorch framework to boost the adversarial transferability for image classification.

GeoThinker. Python

openclaw. Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

dFactory. Easy and Efficient dLLM Fine-Tuning

Stream-DiffVSR. The official repository of paper "Stream-DiffVSR: Low-Latency Streamable Video Super-Resolution via Auto-Regressive Diffusion"

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.

InvSR. Arbitrary-steps Image Super-resolution via Diffusion Inversion (CVPR 2025)

Self-Forcing. Official codebase for "Self Forcing: Bridging Training and Inference in Autoregressive Video Diffusion" (NeurIPS 2025 Spotlight)

realtime-video. Krea Realtime 14B. An open-source realtime AI video model.

ml-sharp. Sharp Monocular View Synthesis in Less Than a Second

ODB-dLLM. Implementation of "Orchestrating Dual-Boundaries: An Arithmetic Intensity Inspired Acceleration Framework for Diffusion Language Models"

d3LLM. [ICML 2026] d3LLM: Ultra-Fast Diffusion LLM 🚀

Geoshield. Official PyTorch implementation of AAAI 2026 paper Geoshield

FreeDave. Free Draft-and-Verification: Toward Lossless Parallel Decoding for Diffusion Large Language Models

DiRL. Python

JiT. PyTorch implementation of JiT https://arxiv.org/abs/2511.13720

SkyReels-V2. SkyReels-V2: Infinite-length Film Generative model

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

GEN3C. [CVPR 2025 Highlight] GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control

hle. Humanity's Last Exam

SMDM. Official PyTorch implementation for ICLR2025 paper "Scaling up Masked Diffusion Models on Text"

LLaDA. Official PyTorch implementation for "Large Language Diffusion Models"

DiffuLLaMA. [ICLR2025] DiffuGPT and DiffuLLaMA: Scaling Diffusion Language Models via Adaptation from Autoregressive Models

Open-dLLM. Open diffusion language model for code generation — releasing pretraining, evaluation, inference, and checkpoints.

MegaDLMs. GPU-optimized framework for training diffusion language models at any scale. The backend of Quokka, Super Data Learners, and OpenMoE 2 training.

dLLM-RL. [ICLR 2026] Official code for TraceRL: Revolutionizing post-training for Diffusion LLMs, powering the SOTA TraDo series.

dInfer. dInfer: An Efficient Inference Framework for Diffusion Language Models

cambrian-s. Cambrian-S: Towards Spatial Supersensing in Video

dinov3-finetune. Testing adaptation of the DINOv2/3 encoders for vision tasks with Low-Rank Adaptation (LoRA)

OmniSafeBench-MM. A Unified Benchmark and Toolbox for Multimodal Jailbreak Attack–Defense Evaluation

WorldScore. Official implementation for WorldScore: A Unified Evaluation Benchmark for World Generation

AdaCache. Code for our ICCV 2025 paper "Adaptive Caching for Faster Video Generation with Diffusion Transformers"

Fast-dLLM. Official implementation of "Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding"

MagCache. The official code for NeurIPS 2025 "MagCache: Fast Video Generation with Magnitude-Aware Cache"

DeepSpeed. DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

DiffSynth-Studio. Enjoy the magic of Diffusion models!

RAE. Official PyTorch Implementation of "Diffusion Transformers with Representation Autoencoders"

FlashWorld. Code for "FlashWorld: High-quality 3D Scene Generation within Seconds" (ICLR 2026 Oral)

4DNeX. 4DNeX: Feed-Forward 4D Generative Modeling Made Easy

Zero-to-Wan. A minimalistic, hackable code base to finetune Wan video generation model

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

LongLive. Long Video Gen Infrastructure

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

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

colar. [NeurIPS 2025] Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains

verl-tool. A version of verl to support diverse tool use [TMLR 2026]

VTool-R1. [ICLR 2026] "VTool-R1: VLMs Learn to Think with Images via Reinforcement Learning on Multimodal Tool Use"

WeKnora. Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.

Oyster. The Oyster series is a set of safety models developed in-house by Alibaba-AAIG, devoted to building a responsible AI ecosystem. | Oyster 系列是 Alibaba-AAIG 自研的安全模型，致力于构建负责任的 AI 生态。

rStar. Python

PREMIR. [EMNLP 2025] The official implementation of "Zero-shot Multimodal Document Retrieval via Cross-Modal Question Generation"

VRAG. Multimodal Retrieval-augmented Generation Framework Built by Tongyi Lab, Alibaba Group.

R-Zero. [ICLR2026] codes for R-Zero: Self-Evolving Reasoning LLM from Zero Data (https://www.arxiv.org/pdf/2508.05004)

Thyme. ✨✨ [ICLR 2026] Think Beyond Images

OmniSpatial. [ICLR 2026] OmniSpatial: Towards Comprehensive Spatial Reasoning Benchmark for Vision Language Models

LVSM. [ICLR 2025 Oral] Official code for "LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias"

UniUGG. [ICLR 2026] UniUGG: Unified 3D Understanding and Generation via Geometric-Semantic Encoding

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

Bagel-Zebra-CoT. https://huggingface.co/datasets/multimodal-reasoning-lab/Zebra-CoT

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

TreeVGR. [ICLR'26] Traceable Evidence Enhanced Visual Grounded Reasoning: Evaluation and Methodology

dots.vlm1. The official repository of the dots.vlm1 instruct models proposed by rednote-hilab.

DeepSeek-R1.

DeepSeek-V3. Python

Aurora-perception. Python

metamorph. Code for MetaMorph Multimodal Understanding and Generation via Instruction Tuning

Director3D. Code for "Director3D: Real-world Camera Trajectory and 3D Scene Generation from Text" (NeurIPS 2024).

UnifiedReward. Official implementation of UnifiedReward & [NeurIPS 2025] UnifiedReward-Think & UnifiedReward-Flex

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

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

Awesome-Multimodal-Large-Language-Models. :sparkles::sparkles:Latest Advances on Multimodal Large Language Models

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

LatentCoT-Horizon. 📖 This is a repository for organizing papers, codes, and other resources related to Latent Reasoning.

Visual-CoT. [Neurips'24 Spotlight] Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoning

YUME. The official code of Yume

reloc3r. [CVPR 2025] Relative camera pose estimation and visual localization with Reloc3r

Response-Attack. Official implementation of “Response Attack: Exploiting Contextual Priming to Jailbreak Large Language Models” (AAAI 2026).

ross. [ICLR'25] Reconstructive Visual Instruction Tuning

Awesome-Unified-Multimodal-Models. Awesome Unified Multimodal Models

bev-vae. BEV-VAE: A Unified BEV Representation for Generalizable Driving Scene Synthesis

gill. 🐟 Code and models for the NeurIPS 2023 paper "Generating Images with Multimodal Language Models".

GoT. Official repository of "GoT: Unleashing Reasoning Capability of Multimodal Large Language Model for Visual Generation and Editing"

MoVieS. [CVPR 2026] Official implementation of "MoVieS: Motion-Aware 4D Dynamic View Synthesis in One Second".

siren. Welcome to the official repository for Siren, a project aimed at understanding and mitigating harmful behaviors in large language models (LLMs). This repository contains the resources for reproducing the experiments described in our work.

GuardReasoner. [ICLR Workshop 2025] An official source code for paper "GuardReasoner: Towards Reasoning-based LLM Safeguards".

Kimi-K2. Kimi K2 is the large language model series developed by Moonshot AI team

Ovis. A novel Multimodal Large Language Model (MLLM) architecture, designed to structurally align visual and textual embeddings.

OmniGen2. OmniGen2: Exploration to Advanced Multimodal Generation. https://arxiv.org/abs/2506.18871

ReasonBrain. 【ICML2026】Reasoning to Edit: Hypothetical Instruction-Based Image Editing with Visual Reasoning

SplatFlow. [CVPR 2025] Official code for the paper "SplatFlow: Multi-View Rectified Flow Model for 3D Gaussian Splatting Synthesis"

Ovis-U1. An unified model that seamlessly integrates multimodal understanding, text-to-image generation, and image editing within a single powerful framework.

x-teaming. Python

Mirage. [CVPR 2026] Machine Mental Imagery: Empower Multimodal Reasoning with Latent Visual Tokens

panda-guard. Panda Guard is designed for researching jailbreak attacks, defenses, and evaluation algorithms for large language models (LLMs).

ActorAttack. Python

RACE. Python

JailTrickBench. Bag of Tricks: Benchmarking of Jailbreak Attacks on LLMs. Empirical tricks for LLM Jailbreaking. (NeurIPS 2024)

IB4LLMs. [NeurIPS'24] Protecting Your LLMs with Information Bottleneck

AutoDefense. AutoDefense: Multi-Agent LLM Defense against Jailbreak Attacks

Agent-Smith. [ICML 2024] Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fast

FlipAttack. [ICML 2025] An official source code for paper "FlipAttack: Jailbreak LLMs via Flipping".

JOOD. [CVPR 2025] Official implementation for JOOD "Playing the Fool: Jailbreaking LLMs and Multimodal LLMs with Out-of-Distribution Strategy"

Awesome-Jailbreak-on-LLMs. Awesome-Jailbreak-on-LLMs is a collection of state-of-the-art, novel, exciting jailbreak methods on LLMs. It contains papers, codes, datasets, evaluations, and analyses.

Vision-Matters. (ArXiv25) Vision Matters: Simple Visual Perturbations Can Boost Multimodal Math Reasoning

LLM-CBRN-Risks. [ACL 2025 Findings] The official GitHub repo for the paper "Nuclear Deployed: Analyzing Catastrophic Risks in Decision-making of Autonomous LLM Agents"

Generalization_unified_VLM. Python

ASVR. Autoregressive Semantic Visual Reconstruction Helps VLMs Understand Better

vjepa2. PyTorch code and models for VJEPA2 self-supervised learning from video.

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

3rgs. Python

AnySplat. [SIGGRAPH Asia 2025 (ACM TOG)] AnySplat: Feed-forward 3D Gaussian Splatting from Unconstrained Views

3dgs-mcmc. [NeurIPS 2024 Spotlight] Implementation of the paper "3D Gaussian Splatting as Markov Chain Monte Carlo"

sparf. This is the official code release for SPARF: Neural Radiance Fields from Sparse and Noisy Poses [CVPR 2023-Highlight]

Bagel. Open-source unified multimodal model

BrickGPT. [ICCV 2025 Best Paper] Official repository for BrickGPT, the first approach for generating physically stable toy brick models from text prompts.

Awesome-3D-Scene-Generation. A curated list of awesome 3D scene generation papers. (arXiv 2505.05474)

Deep-Live-Cam. real time face swap and one-click video deepfake with only a single image

edm2. EDM2 and Autoguidance -- Official PyTorch implementation

EVolSplat. official code of CVPR2025 Evolsplat

AlphaEdit. AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models, ICLR 2025 (Outstanding Paper)

MaskDiT. Code for Fast Training of Diffusion Models with Masked Transformers

FlexWorld. Official PyTorch implementation for "FlexWorld: Progressively Expanding 3D Scenes for Flexiable-View Synthesis".

WorldGen. 🌍 WorldGen - Generate Any 3D Scene in Seconds