Yongdong Luo

AI-Research-SKILLs. Comprehensive open-source library of AI research and engineering skills for any AI model. Package the skills and your claude code/codex/gemini agent will be an AI research agent with full horsepower. Maintained by Orchestra Research.

humanizer. Agent skill that removes signs of AI-generated writing from text

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。

Valley. Valley is a cutting-edge multimodal large model designed to handle a variety of tasks involving text, images, video, and audio data.

UltraEval-Audio. Your faithful, impartial partner for audio evaluation — know yourself, know your rivals. 真实评测，知己知彼。A unified benchmark framework for ASR/TTS/Audio Codec/audio LLM evaluation

AudioBench. AudioBench: A Universal Benchmark for Audio Large Language Models

MELD. MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversation

MMAU. Python

paper-piggy. 🐷 论文小猪 PaperPiggy —— 让 AI agent 真正读懂你的文献库：39 个 MCP 工具 · 引文核对与论断三态核验 · 可追溯原文定位 · 可累积的综合层 Wiki。面向法学/社科研究者。

OmniScope. [ACMMM 2026🔥] This is the official implementation of our paper "OmniScope: Modality-decoupled Token Compression for Efficient Omnimodal Video Understanding"

academic-research-skills. Academic Research Skills for Claude Code: research → write → review → revise → finalize

harness-engineering. Harness Engineering 学习指南 — 从概念理解到独立实践的深度学习档案

future-agi. Open-source, end-to-end platform for evaluating, observing, and improving LLM and AI agent applications. Tracing · Evals · Simulations · Datasets · Gateway · Guardrails. Self-hostable. Apache 2.0.

MindPipe. A powerful model compression framework for LLMs and LVLMs, adapted for NVIDIA GPUs and Huawei Ascend NPUs.

PoCo. [CVPR2026] Official implementation of our paper “Rethinking Position Embedding as a Context Controller for Multi-Reference and Multi-Shot Video Generation”

Video-MME-v2. Video-MME-v2: Towards the Next Stage in Benchmarks for Comprehensive Video Understanding

SpecEyes. [ECCV 2026🔥] This is the official implementation of our paper "SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning"

WFS-SB. [CVPR 2026] Wavelet-based Frame Selection by Detecting Semantic Boundary for Long Video Understanding

HunyuanOCR. HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Better

GOT-OCR2.0. Official code implementation of General OCR Theory: Towards OCR-2.0 via a Unified End-to-end Model

SPEED. [ICLR'26] SPEED: Scalable, Precise, and Efficient Concept Erasure for Diffusion Models

SAFE. [KDD'25] Improving Synthetic Image Detection Towards Generalization: An Image Transformation Perspective

2S-GDA. Python

Video-RAG-Ultra. Multi-modal video understanding system based on RAG technology

MLLM-Token-Compression. Towards Efficient Multimodal Large Language Models: A Survey on Token Compression

IPDN. Python

3D-DRES. Python

Agent-Memory-Paper-List. The paper list of "Memory in the Age of AI Agents: A Survey"

KTS. Kernel Temporal Segmentation

TimeChat. [CVPR 2024] TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding

DDVC. This repo is a official codebase for our paper accepted to the ACL2025. The aim of this repo is to help other researchers

CAT-V. [AAAI 26 Demo] Offical repo for CAT-V - Caption Anything in Video: Object-centric Dense Video Captioning with Spatiotemporal Multimodal Prompting

LongCat-Flash-Omni. This is the official repo for the paper "LongCat-Flash-Omni Technical Report"

HunyuanVideo-1.5. HunyuanVideo-1.5: A leading lightweight video generation model

OmniZip. [CVPR 2026] OmniZip: Audio-Guided Dynamic Token Compression for Fast Omnimodal Large Language Models

VideoMind. 🧠 VideoMind: A Chain-of-LoRA Agent for Temporal-Grounded Video Reasoning (ICLR 2026)

CLAP. Contrastive Language-Audio Pretraining

Ming. Ming - facilitating advanced multimodal understanding and generation capabilities built upon the Ling LLM.

STTM. [ICCV 2025] Multi-Granular Spatio-Temporal Token Merging for Training-Free Acceleration of Video LLMs

HoliTom. [NeurIPS 2025] HoliTom: Holistic Token Merging for Fast Video Large Language Models

VITA. The official implement of VITA, VITA15, LongVITA, VITA-Audio, VITA-VLA, and VITA-E.

Open-o3-Video. [ICML 2026] Official implementation of "Open-o3 Video: Grounded Video Reasoning with Explicit Spatio-Temporal Evidence"

RWKV-LM. RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.

flash-linear-attention. 🚀 Efficient implementations for emerging model architectures

DATE. Use 2 lines to empower absolute time awareness for Qwen2.5VL's MRoPE

VidCom2. [EMNLP 2025 Main] Video Compression Commander: Plug-and-Play Inference Acceleration for Video Large Language Models

VFlowOpt. [ICCV 2025] VFlowOpt: A Token Pruning Framework for LMMs with Visual Information Flow-Guided Optimization

ReAgent-V. [NeurIPS'25] ReAgent-V: A Reward-Driven Multi-Agent Framework for Video Understanding

mini-swe-agent. The 100 line AI agent that solves GitHub issues or helps you in your command line. Radically simple, no huge configs, no giant monorepo—but scores >74% on SWE-bench verified!

Awesome-Multimodal-Token-Compression. [TMLR 2026] Survey: https://arxiv.org/pdf/2507.20198

Qwen3-Omni. 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.

Project-Imaging-X. Project Imaging-X: A Survey of 1000+ Open-Access Medical Imaging Datasets for Foundation Model Development

DHAT. [ICCV 2025] Towards Adversarial Robustness via Debiased High-Confidence Logit Alignment

LLaVA-OneVision-2. Fully Open Framework for Democratized Multimodal Training

mcp_chatbot. A chatbot implementation compatible with MCP (terminal / streamlit supported)

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

T2I-CoReBench. [ICLR'26] Easier Painting Than Thinking: Can Text-to-Image Models Set the Stage, but Not Direct the Play?

minimind. 🧠「大模型」2小时完全从0训练64M的小参数LLM！Train a 64M-parameter LLM from scratch in just 2h!

awesome-nano-banana. Awesome curated collection of images and prompts generated by gemini-2.5-flash-image (aka Nano Banana) state-of-the-art image generation and editing model. Explore AI generated visuals created with Gemini, showcasing Google’s advanced image generation capabilities.

youtu-agent. A simple yet powerful agent framework that delivers with open-source models

ViaRL. ViaRL: Adaptive Temporal Grounding via Visual Iterated Amplification Reinforcement Learning

ICoT. [CVPR' 25] Interleaved-Modal Chain-of-Thought

Awesome-MCoT. Multimodal Chain-of-Thought Reasoning: A Comprehensive Survey

MINT-CoT. [NeurIPS 2025] MINT-CoT: Enabling Interleaved Visual Tokens in Mathematical Chain-of-Thought Reasoning

gpt-oss. gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI

ARC-Hunyuan-Video-7B. Structured Video Comprehension of Real-World Shorts

LLoVi. Official implementation for "A Simple LLM Framework for Long-Range Video Question-Answering"

VideoAgent. Python

BOLT. [CVPR2025] BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding

AKS. [CVPR 2025] Adaptive Keyframe Sampling for Long Video Understanding

unsloth-zoo. Utils for Unsloth https://github.com/unslothai/unsloth

verl. verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework

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

flash-attention. Fast and memory-efficient exact attention

BackMix. [TPAMI2025] BackMix: Regularizing Open Set Recognition by Removing Underlying Fore-Background Priors

MEDAF. [AAAI2024] Exploring Diverse Representations for Open Set Recognition

Liger-Kernel. Efficient Triton Kernels for LLM Training

MMPerspective. [NeurIPS 2025] A Comprehensive Benchmark for Perspective Perception, Reasoning, and Robustness

Do-You-See-Me. Python

RTV-Bench. [NeurIPS 2025] 𝓡𝓣𝓥-𝓑𝓮𝓷𝓬𝓱: Benchmarking MLLM Continuous Perception, Understanding and Reasoning through Real-Time Video.

ms-swift. Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025).

VideoMathQA. VideoMathQA is a benchmark designed to evaluate mathematical reasoning in real-world educational videos

VidText. Comprehensive benchmark for video text understanding

OVO-Bench. [CVPR 2025] OVO-Bench: How Far is Your Video-LLMs from Real-World Online Video Understanding?

SLOT. Python

Keye. Python

ROLL. An Efficient and User-Friendly Scaling Library for Reinforcement Learning with Large Language Models

MM-EUREKA. MM-EUREKA: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning

AdaVideoRAG. [NeurIPS 2025] AdaVideoRAG: Omni-Contextual Adaptive Retrieval-Augmented Efficient Long Video Understanding

VisualQuality-R1. [NeurIPS 2025 Spotlight] VisualQuality-R1 is the first open-sourced NR-IQA model can accurately describe and rate the image quality.

Awesome-Image-Quality-Assessment. A comprehensive collection of IQA papers

oss-browser. OSS Browser 提供类似windows资源管理器功能。用户可以很方便的浏览文件，上传下载文件，支持断点续传等。

Co-Instruct. ④[ECCV 2024 Oral, Comparison among Multiple Images!] A study on open-ended multi-image quality comparison: a dataset, a model and a benchmark.

SILVR. Official Implementation for "SiLVR : A Simple Language-based Video Reasoning Framework"

AI-Agent-papers. Collection of recent works on AI Agents.

unsloth. Unsloth is a local UI for training and running Kimi K3, Gemma 4, Qwen3.6, DeepSeek, GLM and other models.

Label-Free-RLVR.

Temporal-R1. Reinforcement Learning Tuning for VideoLLMs: Reward Design and Data Efficiency

VAP. [NeurIPS 2025] Poison as Cure: Visual Noise for Mitigating Object Hallucinations in LVMs

Q-Insight. Q-Insight Family: Q-Insight, VQ-Insight and RALI (NeurIPS 2025 Spotlight, AAAI 2026 Oral, and ICLR 2026 Oral)

Video-Holmes. [ECCV 2026] Video-Holmes: Can MLLM Think Like Holmes for Complex Video Reasoning?

Omni-R1. [NeurIPS 2025] Official Repo of Omni-R1: Reinforcement Learning for Omnimodal Reasoning via Two-System Collaboration

LlamaFactory. Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)

Pixel-Reasoner. Pixel-Level Reasoning Model trained with RL [NeuIPS25]

Paper-Writing-Tips. MLNLP社区用来帮助大家避免论文投稿小错误的整理仓库。 Paper Writing Tips

Sora-Generates-Videos-with-Stunning-Geometrical-Consistency. Sora Generates Videos with Stunning Geometrical Consistency

AIGCBench. [TBench 2024] Official implementation of "AIGCBench: Comprehensive Evaluation of Image-to-Video Content Generated by AI"

Minimal-RL. Python

KuaiMod.github.io. JavaScript

Qwen-VL-Series-Finetune. An open-source implementaion for fine-tuning Qwen-VL series by Alibaba Cloud.

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

Q-Insight. Q-Insight is open-sourced at https://github.com/bytedance/Q-Insight. This repository will not receive further updates.

AnchorCrafter. Python

VITA-Audio. ✨✨[NeurIPS 2025] VITA-Audio: Fast Interleaved Cross-Modal Token Generation for Efficient Large Speech-Language Model

trl. Train transformer language models with reinforcement learning.

open-r1. Fully open reproduction of DeepSeek-R1

VideoChat-R1. [NIPS2025] VideoChat-R1 & R1.5: Enhancing Spatio-Temporal Perception and Reasoning via Reinforcement Fine-Tuning

HowToCook. Programmer's guide about how to cook at home.

FRAG. Python

MR-Video. MR. Video: MapReduce is the Principle for Long Video Understanding

OpenRLHF. An Easy-to-use, Scalable and High-performance Agentic RL Framework based on Ray (PPO & DAPO & REINFORCE++ & VLM & TIS & vLLM & Ray & Async RL)

Video-MMLU. A Massive Multi-Discipline Lecture Understanding Benchmark

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

ViLAMP. [ICML 2025] Official repository for paper "Scaling Video-Language Models to 10K Frames via Hierarchical Differential Distillation"

describe-anything. [ICCV 2025] Implementation for Describe Anything: Detailed Localized Image and Video Captioning

VidCapBench. Python

tarsier. Tarsier -- a family of large-scale video-language models, which is designed to generate high-quality video descriptions , together with good capability of general video understanding.

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

VPO. Python

VisionReward. [AAAI 2026] VisionReward: Fine-Grained Multi-Dimensional Human Preference Learning for Image and Video Generation

HumanAesExpert. Official implementation of "HumanAesExpert: Advancing a Multi-Modality Foundation Model for Human Image Aesthetic Assessment"

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

tensorzero. TensorZero is an open-source LLMOps platform that unifies an LLM gateway, observability, evaluation, optimization, and experimentation.

SymbolicDet. Python

VideoAlign. [NeurIPS 2025] Improving Video Generation with Human Feedback

vllm. A high-throughput and memory-efficient inference and serving engine for LLMs

Qwen2.5-Omni. Qwen2.5-Omni is an end-to-end multimodal model by Qwen team at Alibaba Cloud, capable of understanding text, audio, vision, video, and performing real-time speech generation.

Time-R1. R1-like Video-LLM for Temporal Grounding

aurora. [ICLR 2025] AuroraCap: Efficient, Performant Video Detailed Captioning and a New Benchmark

SlideChat. SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding

Campus2026. 2026届互联网校招&2025互联网实习信息汇总，欢迎共创

PE3R. [CVPR'26] PE3R: Perception-Efficient 3D Reconstruction. Take 2 - 3 photos with your phone, upload them, wait a few minutes, and then start exploring your 3D world via text!

Sparrow. Sparrow: Data-Efficient Video-LLM with Text-to-Image Augmentation

Vision-R1. [ICLR2026] This is the first paper to explore how to effectively use R1-like RL for MLLMs and introduce Vision-R1, a reasoning MLLM that leverages cold-start initialization and RL training to incentivize reasoning capability.

E3-FaceNet. [ICML 2024] Fast Text-to-3D-Aware Face Generation and Manipulation via Direct Cross-modal Mapping and Geometric Regularization

StoryWeaver. [AAAI 2025] StoryWeaver: A Unified World Model for Knowledge-Enhanced Story Character Customization

ACL25-AdaReTaKe. Official implementation of paper AdaReTaKe: Adaptive Redundancy Reduction to Perceive Longer for Video-language Understanding

QuoTA. ✨✨[AAAI 2026] This is the official implementation of our paper "QuoTA: Query-oriented Token Assignment via CoT Query Decouple for Long Video Comprehension"

TPO. Python

Woodpecker. ✨✨Woodpecker: Hallucination Correction for Multimodal Large Language Models

LinVT. LinVT: Empower Your Image-level Large Language Model to Understand Videos

VideoLLaMA3. Frontier Multimodal Foundation Models for Image and Video Understanding

VideoNIAH. VideoNIAH: A Flexible Synthetic Method for Benchmarking Video MLLMs

VILA. VILA is a family of state-of-the-art vision language models (VLMs) for diverse multimodal AI tasks across the edge, data center, and cloud.

VideoChat-Flash. [ICLR2026] VideoChat-Flash: Hierarchical Compression for Long-Context Video Modeling

Evolver. [COLING 2025🔥] Evolver: Chain-of-Evolution Prompting to Boost Large Multimodal Models for Hateful Meme Detection

LongVU. [ICML 2025] Official PyTorch implementation of LongVU

FrameFusion. [ICCV'25] The official code of paper "Combining Similarity and Importance for Video Token Reduction on Large Visual Language Models"

FastV. [ECCV 2024 Oral] Code for paper: An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models

DyCoke. [CVPR 2025] DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Models

Awesome-Token-Compress. A paper list of some recent works about Token Compress for Vit and VLM

DeepSeek-R1.

Popkart-Excess-system-modelling. Python

GaitRDAE. Official Repository of GaitRDAE.

LVBench. [ICCV 2025] LVBench: An Extreme Long Video Understanding Benchmark

NExT-QA. NExT-QA: Next Phase of Question-Answering to Explaining Temporal Actions (CVPR'21)

MiniCPM-V. A Pocket-Sized MLLM for Ultra-Efficient Image and Video Understanding on Your Phone

video-ReTaKe. Official implementation of paper ReTaKe: Reducing Temporal and Knowledge Redundancy for Long Video Understanding

Thinking-Claude. Let your Claude able to think

MBQ. The code repository of "MBQ: Modality-Balanced Quantization for Large Vision-Language Models"

ToMe. A method to increase the speed and lower the memory footprint of existing vision transformers.

AIM. [ICCV 2025] Official code for "AIM: Adaptive Inference of Multi-Modal LLMs via Token Merging and Pruning"

TAR3D. Official Code for 'TAR3D: Creating High-Quality 3D Assets via Next-Part Prediction' (ICCV 2025)

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

MMStar. [NeurIPS 2024] This repo contains evaluation code for the paper "Are We on the Right Way for Evaluating Large Vision-Language Models"

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

MDIN. [MM2024 Oral] 3D-GRES: Generalized 3D Referring Expression Segmentation

RG-SAN. [NeurIPS 2024 Oral] RG-SAN: Rule-Guided Spatial Awareness Network for End-to-End 3D Referring Expression Segmentation

Sparrow. Repo for paper "T2Vid: Translating Long Text into Multi-Image is the Catalyst for Video-LLMs"

AutoGPTQ. An easy-to-use LLMs quantization package with user-friendly apis, based on GPTQ algorithm.

bitsandbytes. Accessible large language models via k-bit quantization for PyTorch.

SAVEn-Vid. SAVEn-Vid: Synergistic Audio-Visual Integration for Enhanced Understanding in Long Video Context

QVLM. [NeurIPS'24]Efficient and accurate memory saving method towards W4A4 large multi-modal models.

video-rag.github.io. JavaScript

Video-RAG-master. ✨✨[NeurIPS 2025] This is the official implementation of our paper "Video-RAG: Visually-aligned Retrieval-Augmented Long Video Comprehension"

Freeze-Omni. ✨✨Freeze-Omni: A Smart and Low Latency Speech-to-speech Dialogue Model with Frozen LLM

Video-XL. 🔥🔥First-ever hour scale video understanding models

LongVideoBench. [Neurips 24' D&B] Official Dataloader and Evaluation Scripts for LongVideoBench.

MLVU. 🔥🔥MLVU: Multi-task Long Video Understanding Benchmark

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

VideoAgent. This is the official code of VideoAgent: A Memory-augmented Multimodal Agent for Video Understanding (ECCV 2024)

peiqianjichang. 赔钱机场官网地址

EgoSchema. Python

lmms-eval. One-for-All Multimodal Evaluation Toolkit Across Text, Image, Video, and Audio Tasks

DPR. Dense Passage Retriever - is a set of tools and models for open domain Q&A task.

3DGCTR. This is a PyTorch implementation of 3DGCTR proposed by our paper “Rethinking 3D Dense Caption and Visual Grounding in A Unified Framework through Prompt-based Localization”

ircot. Repository for Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions, ACL23

mini-omni. open-source multimodal large language model that can hear, talk while thinking. Featuring real-time end-to-end speech input and streaming audio output conversational capabilities.

YoLLaVA. 🌋👵🏻 Yo'LLaVA: Your Personalized Language and Vision Assistant (NeurIPS 2024)

video-mme.github.io. JavaScript