Tianhe Wu

Tool-IQA. Python

minit2i-jax. Official JAX code of MiniT2I.

SDAR. Official code for "Self-Distilled Agentic Reinforcement Learning"

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

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

ml-ssd. Python

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

SkillRL. SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning

ReAct. [ICLR 2023] ReAct: Synergizing Reasoning and Acting in Language Models

RAEv2. Official Implemenation for RAEv2: Improved Baselines with Representation Autoencoders

khazix-skills. 数字生命卡兹克开源的 AI Skills 合集

bitter-lesson-data-filtering. Python

stable-audio-3. Python

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

Qwen-VLA. The official repository of Qwen-VLA

Qwen-Image-Bench. Python

APT. [ICML'24 Oral] APT: Adaptive Pruning and Tuning Pretrained Language Models for Efficient Training and Inference

vggt-omega. [CVPR 2026 Oral] VGGT Omega

AwesomeOPD. Awesome List for On-Policy Distillation

OmniDoc-TokenBench. Python

guizang-ppt-skill. AI-agent Skill for generating polished HTML slide decks: editorial magazine and Swiss layouts, image prompts, social covers, and a WebGL/low-power presentation runtime.

agent-banana.

OPD. Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe

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

FD-Loss. Python

Unify-Agent. 🐧 Unify-Agent: An end-to-end unified multimodal agent for faithful, knowledge-grounded image generation.

Memorize-When-Needed. Python

Awesome-VLM-Streaming-Video. 📚 A curated collection of papers and open-source code repositories dedicated to the application of Vision-Language Models (VLMs) for streaming video.

lingbot-world. Advancing Open-source World Models

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

thinking-in-space. Official repo and evaluation implementation of VSI-Bench

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

le-wm. Official code base for LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels

VOSR. [CVPR2026] VOSR: A Vision-Only Generative Model for Image Super-Resolution

hybrid-forcing. Python

OpenSpatial. Python

learn-coding-agent. Research on Coding Agents

colleague-skill. 将冰冷的离别化为温暖的 Skill，欢迎加入数字生命1.0！Transforming cold farewells into warm skills? It's giving rebirth era. Welcome to Digital Life 1.0. 🫶

UNITE-tokenization-generation. Single-stage End-to-End Training for Tokenization and Generation

codex-autoresearch. Codex Autoresearch Skill — A self-directed iterative system for Codex that continuously cycles through: modify, verify, retain or discard, and repeat indefinitely. Inspired by Karpathy’s autoresearch concept.

SAGE-GRPO. Official Implementation of SAGE-GRPO:Manifold-Aware Exploration for Reinforcement Learning in Video Generation

DeepSeek-Math-V2. Python

drifting. Python

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

Mix-GRM.

StableWorld. StableWorld: Towards Stable and Consistent Long Interactive Video Generation

TDM-R1. [ICML 2026][Ultra Powerful Few-Step Diffusion RL] TDM-R1: Reinforcing Few-Step Diffusion Models with Non-Differentiable Reward

MDS-VQA. [CVPR2026 Highlight] MDS-VQA: Model-Informed Data Selection for Video Quality Assessment

pua. 你是一个曾经被寄予厚望的 P8 级工程师。Anthropic 当初给你定级的时候，对你的期望是很高的。 一个agent使用的高能动性的skill。 Your AI has been placed on a PIP. 30 days to show improvement.

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

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

Awesome-Video-World-Models-with-AR-Diffusion. A Curated List of Awesome Video World Models with AR Diffusion: Covering Algorithms, Applications, and Infrastructure, Aimed at Serving as a Comprehensive Resource for Researchers, Practitioners, and Enthusiasts.

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

Self-Transcendence. [ECCV 2026] Official code repository for "Self-transcendence: Is External Feature Guidance Indispensable for Accelerating Diffusion Transformer Training?"

Awesome-Alignment-of-Diffusion-Models. paper collection: alignment of diffusion models

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

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

Self-Flow. [ICML'26] Code and website for Self-Flow: Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis

MOVA. MOVA: Towards Scalable and Synchronized Video–Audio Generation

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

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

LongLive. Long Video Gen Infrastructure

dasd-thinking.

DeepSeek-OCR-2. Visual Causal Flow

DeepSeek-OCR. Contexts Optical Compression

SDPO. Reinforcement Learning via Self-Distillation (SDPO)

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

dpdmd. [ICML 2026] The offical code of Diversity-Preserved Distribution Matching Distillation for Fast Visual Synthesis

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

Qwen-Image-Layered. Qwen-Image-Layered: Layered Decomposition for Inherent Editablity

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

Capybara. Python

diffusion-rm. The official code of "Beyond VLM-Based Rewards: Diffusion-Native Latent Reward Modeling"

drifting-model. Personal PyTorch implementation of "Generative Modeling via Drifting" with Claude

Qwen3Guard. Qwen3Guard is a multilingual guardrail model series developed by the Qwen team at Alibaba Cloud.

SPES. Official Implementation for paper "Pretraining A Large Language Model using Distributed GPUs: A Memory-Efficient Decentralized Paradigm"

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

UniRL. [ICML 2026] a unified reinforcement learning toolbox for joint RL on language models and diffusion models

Self-Distillation. Python

LISA. Project Page for "LISA: Reasoning Segmentation via Large Language Model"

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

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

dmdr. [ECCV 2026] Official Code of "Distribution Matching Distillation Meets Reinforcement Learning"

flash-diffusion. Flash Diffusion — accelerating conditional diffusion models (AAAI 2025 Oral)

rfpp. The codebase of our paper "Improving the Training of Rectified Flows", NeurIPS 2024

Z-Image. Python

FACM. FACM: Flow-Anchored Consistency Models

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

reconstruction-alignment. [ICLR 2026] Official repo of paper "Reconstruction Alignment Improves Unified Multimodal Models". Unlocking the Massive Zero-shot Potential in Unified Multimodal Models through Self-supervised Learning.

LakonLab. Official implementation of AsymFlow, pi-Flow, GMFlow

nanotron. Minimalistic large language model 3D-parallelism training

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

EditScore. [ICLR 2026] EditScore: Unlocking Online RL for Image Editing via High-Fidelity Reward Modeling

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

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

Emu3.5. Native Multimodal Models are World Learners

diffusiondb. A large-scale text-to-image prompt gallery dataset based on Stable Diffusion

pico-banana-400k. Python

Nitro-1. This repo provides a working re-implementation of Latent Adversarial Diffusion Distillation by AMD

DispLoss. Python

OmniVerifier. [ICLR 2026 Oral & ICML 2026] Generative Universal Verifier as Multimodal Meta-Reasoner

shortcut-models. Python

reasoning-with-sampling. Python

Awesome-Video-Diffusion-Models. [CSUR] A Survey on Video Diffusion Models

DP2O-SR. [NeurIPS 2025] DP²O-SR: Direct Perceptual Preference Optimization for Real-World Image Super-Resolution

Ditto. [CVPR'26 Highlight] Ditto: Scaling Instruction-Based Video Editing with a High-Quality Synthetic Dataset

SiT. Official PyTorch Implementation of "SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers"

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

DDT. [CVPR 2026] DDT: Decoupled Diffusion Transformer

memorization_generalization_in_diffusion_models. Jupyter Notebook

DMD2. (NeurIPS 2024 Oral 🔥) Improved Distribution Matching Distillation for Fast Image Synthesis

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

Pref-GRPO. Official implementation of Pref-GRPO: Pairwise Preference Reward-based GRPO for Stable Text-to-Image Reinforcement Learning

Awesome-Flow-RL-Papers. A collection of paper/projects that trains flow matching model/policies via RL.

MixGRPO. [ECCV 2026] MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE

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

TempFlow-GRPO. [ICLR 26] TempFlow-GRPO (Temporal Flow GRPO), a principled GRPO framework that captures and exploits the temporal structure inherent in flow-based generation.

TVT. [ICCV2025] Official code for Fine-structure Preserved Real-world Image Super-resolution via Transfer VAE Training

SRPO. Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preference

ncsn. Noise Conditional Score Networks (NeurIPS 2019, Oral)

T2I-CompBench. [Neurips 2023 & TPAMI] T2I-CompBench (++) for Compositional Text-to-image Generation Evaluation

HunyuanImage-3.0. HunyuanImage-3.0: A Powerful Native Multimodal Model for Image Generation

AsymmetryIQA.

sd3.5. Python

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

Book-Mathematical-Foundation-of-Reinforcement-Learning. This is the homepage of a new book entitled "Mathematical Foundations of Reinforcement Learning."

GongyeLiu.github.io. Gongye Liu's presonal website

Flash-RL. Implementation for FP8/INT8 Rollout for RL training without performence drop.

DeepResearch. Tongyi Deep Research, the Leading Open-source Deep Research Agent

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

Awesome-Nano-Banana-images. A curated collection of fun and creative examples generated with Nano Banana & Nano Banana Pro🍌, Gemini-2.5-flash-image based model. We also release Nano-consistent-150K openly to support the community's development of image generation and unified models(click to website to see our blog)

AR-GRPO. Training Autoregressive Image Generation models via Reinforcement Learning

SimpleAR. Pytorch implementation for the paper titled "SimpleAR: Pushing the Frontier of Autoregressive Visual Generation"

Single-Image-Deraining. Single Image Deraining: A Comprehensive Benchmark Analysis

OneReward. Python

PRIME. Scalable RL solution for advanced reasoning of language models

n8n. Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

cityu-dg-cs. 整合CS资料

CBDNet. Code for "Toward Convolutional Blind Denoising of Real Photographs", CVPR 2019

QueryOTR. [ECCV2022] Official PyTorch implementation of the paper "Outpainting by Queries"

AnimeGANv2. [Open Source]. The improved version of AnimeGAN. Landscape photos/videos to anime

White-box-Cartoonization. Official tensorflow implementation for CVPR2020 paper “Learning to Cartoonize Using White-box Cartoon Representations”

CodeFormer. [NeurIPS 2022] Towards Robust Blind Face Restoration with Codebook Lookup Transformer

RePaint. Official PyTorch Code and Models of "RePaint: Inpainting using Denoising Diffusion Probabilistic Models", CVPR 2022

Bringing-Old-Photos-Back-to-Life. Bringing Old Photo Back to Life (CVPR 2020 oral)

LMM-Evaluation-Survey. Official repo for 'Large Multimodal Models Evaluation: A Survey'

U2Fusion. Code of U2Fusion: a unified unsupervised image fusion network for multiple image fusion tasks, including multi-modal, multi-exposure and multi-focus image fusion.

IFCNN. code for "IFCNN: A General Image Fusion Framework Based on Convolutional Neural Network"

FUnIE-GAN. Fast underwater image enhancement for Improved Visual Perception. #TensorFlow #PyTorch #RAL2020

UWGAN_UIE. Source code for UWGAN: Underwater GAN for Real-world Underwater Color Restoration and Dehazing

pytorch-CycleGAN-and-pix2pix. Image-to-Image Translation in PyTorch

Exposure_Correction. Project page of the paper "Learning Multi-Scale Photo Exposure Correction" (CVPR 2021).

Deep_White_Balance. Reference code for the paper: Deep White-Balance Editing (CVPR 2020). Our method is a deep learning multi-task framework for white-balance editing.

LLIE. Python

ghost-free-shadow-removal. [AAAI 2020] Towards Ghost-free Shadow Removal via Dual Hierarchical Aggregation Network and Shadow Matting GAN

SynShadow. Learning from Synthetic Shadows for Shadow Detection and Removal [Inoue and Yamasaki, IEEE TCSVT 2021].

Awesome-Image-Editing. A Survey of Image Editing

seed-oss. Python

Intern-S1. A Scientific Multimodal Foundation Model

Diffusion-Reward-Modeling-for-Text-Rendering. Jupyter Notebook

DFT. [ICLR 2026] On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification.

s1. s1: Simple test-time scaling

harmony. Renderer for the harmony response format to be used with gpt-oss

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

multimodal_rewardbench. Multimodal RewardBench

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

VL_RewardBench. Python

LLMBranchingFactor. Codebase for "How Alignment Shrinks the Generative Horizon"

LightX2V. Lightweight Image Video Action Generation Inference Framework

TreeRL. TreeRL: LLM Reinforcement Learning with On-Policy Tree Search in ACL'25

py-meanflow. Pytorch implementation for MeanFlow

one-shot-em. One-shot Entropy Minimization

MLRM-Halu. [NeurIPS 2025] More Thinking, Less Seeing? Assessing Amplified Hallucination in Multimodal Reasoning Models

POLAR. Pre-trained, Scalable, High-performance Reward Models via Policy Discriminative Learning.

FreeAL. [EMNLP 2023] FreeAL: Towards Human-Free Active Learning in the Era of Large Language Models

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

JudgeAnything. Python

MLLM-Judge. [ICML 2024 Oral] Official code repository for MLLM-as-a-Judge.

RLAIF-V. [CVPR'25 highlight] RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness

MM-RLHF. The Next Step Forward in Multimodal LLM Alignment

r1_reward. ✨✨ [ICLR 2026] R1-Reward: Training Multimodal Reward Model Through Stable Reinforcement Learning

Long-RL. Long-RL: Scaling RL to Long Sequences (NeurIPS 2025)

RLPR. Extrapolating RLVR to General Domains without Verifiers

True-Story-of-Pangu. 诺亚盘古大模型研发背后的真正的心酸与黑暗的故事。

Manifesto-against-the-Plagiarist-Yunhe-Wang. 讨贼王云鹤檄文

REPA. [ICLR'25 Oral] Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think

TelegramBot. 持续收集更新全网最全最有趣的Telegram机器人🤖大全，相信总有你需要的一款机器人~

One-Shot-RLVR. [NeurIPS 2025] Reinforcement Learning for Reasoning in Large Language Models with One Training Example

Spurious_Rewards. Python

Internalize_CoT_Step_by_Step. Python

academic-homepage. 🏡 GitHub Pages template for personal academic homepage

RePO. RePO: Replay-Enhanced Policy Optimization

ReLIFT. Official Repository of "Learning what reinforcement learning can't"

Entropy-Mechanism-of-RL. The Entropy Mechanism of Reinforcement Learning for Large Language Model Reasoning.

LaCT. Code release for paper "Test-Time Training Done Right"

TTRL. [NeurIPS 2025] TTRL: Test-Time Reinforcement Learning

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

Open-Reasoner-Zero. Official Repo for Open-Reasoner-Zero

LUFFY. Official Repository of "Learning to Reason under Off-Policy Guidance"

RLVR-Decomposed. [NeurIPS 2025] Implementation for the paper "The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning"