Tong He

cosmos-framework. Our inference and training framework to run on the Cosmos Models

Open-d4rt. Python

SAM2Matting. [ECCV 2026] SAM2Matting: Generalized Image and Video Matting

axolotl. Go ahead and axolotl questions

claude-review-loop. Claude Code plugin: automated code review loop with Codex

i1. Code release for "i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models"

VLX-Flow. VLX-Flow: streaming VLM for real-time general vision intelligence

mmbench2. Official code repository for the paper "Hallucination in World Models is Predictable and Preventable".

Boogu-Image. Boogu-Image-0.1 is an Apache-2.0 open-source image generation and editing model family that delivers near-closed-source performance with an order of magnitude less data.

train-llm-from-scratch. A straightforward method for training your LLM, from downloading data to generating text.

ponytail. Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.

minit2i-torch. Official PyTorch re-implementation of MiniT2I.

JoyAI-VL-Interaction. Python

HRM-Text. HRM-Text is a 1B text generation model based on the HRM architecture, strengthened by task completion and latent space reasoning.

SpatialClaw. SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning

AI-Infra-Auto-Driven-SKILLS. Python

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

humanize. From Automated Idea Factory to Realization

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

Bernini. Bernini is a unified framework for video generation and editing that combines an MLLM-based semantic planner with a DiT-based renderer.

VLM3. Official implementation of paper "VLM³: Vision Language Models Are Native 3D Learners".

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

multica. The open-source managed agents platform. Turn coding agents into real teammates — assign tasks, track progress, compound skills.

NEO. NEO Series: Native Vision-Language Models from First Principles

GeoAlign. Geo-Align: Video Generation Alignment via Metric Geometry Reward

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

OSCAR. OSCAR: Offline Spectral Covariance-Aware Rotation for 2-bit KV Cache Quantization

Lance. A 3B-active-parameter native unified multimodal model for image and video understanding, generation, and editing.

vggt-omega. [CVPR 2026 Oral] VGGT Omega

Megatron-Bridge. Training library for Megatron-based models with bidirectional Hugging Face conversion capability

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

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

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

mammothmoda. Python

Evolving-Visual-Generation. [Roadmap] Visual Generation in the New Era: An Evolution from Atomic Mapping to Agentic World Modeling

MiniCPM. MiniCPM5-1B: A SOTA 1B on-device LLM, small yet powerful.

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.

OpenMythos. A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.

EgoVerse. EgoVerse: Egocentric Data for Robot Learning from Around the World

Spider_XHS. 小红书爬虫数据采集，小红书全域运营解决方案

WildDet3D. Allen Institute for AI: WildDet3D: Scaling Promptable 3D Detection in the Wild

MegaTrain. Python

Spatial-SSRL. [CVPR 2026] Official release of "Spatial-SSRL: Enhancing Spatial Understanding via Self-Supervised Reinforcement Learning"

opencode. The open source coding agent.

vero. Vero: An Open RL Recipe for General Visual Reasoning

mcore-bridge. MCore-Bridge: Providing Megatron-Core model definitions for state-of-the-art large models and making Megatron training as simple as Transformers — with support for 300+ large language models (Qwen3-Next, GLM-5.1, Deepseek-V4, MiniMax-2.7, ...) and 200+ multimodal large models (Qwen3.5, Qwen3-Omni, Gemma4, ...).

eidetic-memory. Own eidetic-memory like Sheldon

cc-connect. Bridge local AI coding agents (Claude Code, Cursor, Gemini CLI, Codex) to messaging platforms (Feishu/Lark, DingTalk, Slack, Telegram, Discord, LINE, WeChat Work). Chat with your AI dev assistant from anywhere — no public IP required for most platforms.

pytube. Lightweight, dependency-free Python library and CLI for downloading YouTube videos, playlists, and captions.

SimpleStream. A simple video streaming baseline that outperforms SOTAs.

happy. Mobile and Web client for Codex and Claude Code, with realtime voice, encryption and fully featured

codex-plugin-cc. Use Codex from Claude Code to review code or delegate tasks.

oh-my-codex. OmX - Oh My codeX: Your codex is not alone. Add hooks, agent teams, HUDs, and so much more.

claw-code. An agent-managed museum exhibit, built in Rust with Gajae-Code / LazyCodex — developed and maintained with no human intervention.

AnyHand. AnyHand: A Large-Scale Synthetic Dataset for RGB(-D) Hand Pose Estimation

DreamID-Omni. [ICML 2026] DreamID-Omni: Unified Framework for Controllable Human-Centric Audio-Video Generation

MagiCompiler. A plug-and-play compiler that delivers free-lunch optimizations for both inference and training.

VideoWorld. [CVPR 2025] VideoWorld is a simple generative model that learns purely from unlabeled videos—much like how babies learn by observing their environment.

HyperAgents. Self-referential self-improving agents that can optimize for any computable task

daVinci-MagiHuman. Python

firecrawl. The API to search, scrape, and interact with the web at scale. 🔥

gstack. Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA

next-ai-draw-io. A next.js web application that integrates AI capabilities with draw.io diagrams. This app allows you to create, modify, and enhance diagrams through natural language commands and AI-assisted visualization.

cowtracker. CoWTracker: Tracking by Warping instead of Correlation

deer-flow. An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.

claude-hud. A Claude Code plugin that shows what's happening - context usage, active tools, running agents, and todo progress

frontend-slides. Create beautiful slides on the web using a coding agent's frontend skills

MemFlow. Official Implementation of "MemFlow: Flowing Adaptive Memory for Consistent and Efficient Long Video Narratives"

AutoResearchClaw. Fully autonomous & self-evolving research from idea to paper. Chat an Idea. Get a Paper. 🦞

DexLatent. [CVPR 2026 Highlight] XL-VLA: Cross-Hand Latent Representation for Vision-Language-Action Models

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.

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

LoGeR. Reimplementation of LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory

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

autoresearch. AI agents running research on single-GPU nanochat training automatically

claude-code-best-practice. from vibe coding to agentic engineering - practice makes claude perfect

Track4World. [ECCV 2026] Track4World: Feedforward World-centric Dense 3D Tracking of All Pixels

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

oh-my-claudecode. Teams-first Multi-agent orchestration for Claude Code

yt-dlp. A feature-rich command-line audio/video downloader

data-juicer. Data processing for and with foundation models! 🍎 🍋 🌽 ➡️ ➡️🍸 🍹 🍷

superset. Code Editor for the AI Agents Era - Run an army of Claude Code, Codex, etc. on your machine

skills. Give your agents the power of the Hugging Face ecosystem

BitDance. BitDance & UniWeTok: Open-source autoregressive model with binary visual tokens. A research project for building powerful multimodal autoregressive model.

nanoclaw. A lightweight alternative to OpenClaw that runs in containers for security. Connects to WhatsApp, Telegram, Slack, Discord, Gmail and other messaging apps,, has memory, scheduled jobs, and runs directly on Anthropic's Agents SDK

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

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

ml-egodex. EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Video

AR-Omni. "AR-Omni: A Unified Autoregressive Model for Any-to-Any Generation"

self-refine-video. [ICML 2026] Pytorch implementation of Self-Refining Video Sampling

superpowers. An agentic skills framework & software development methodology that works.

ECC. The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

vdpm. Official implementation of Video-DPM

Action100M. A Large-scale Video Action Dataset

lance. Open Lakehouse Format for Multimodal AI. Convert from Parquet in 2 lines of code for 100x faster random access, vector index, and data versioning. Compatible with Pandas, DuckDB, Polars, Pyarrow, and PyTorch with more integrations coming..

repomix. 📦 Repomix is a powerful tool that packs your entire repository into a single, AI-friendly file. Perfect for when you need to feed your codebase to Large Language Models (LLMs) or other AI tools like Claude, ChatGPT, DeepSeek, Perplexity, Gemini, Gemma, Llama, Grok, and more.

nanocode. Minimal Claude Code alternative. Single Python file, zero dependencies, ~250 lines.

learn-claude-code. Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1

LTX-2. Official Python inference and LoRA trainer package for the LTX-2 audio–video generative model.

InfiniDepth. [CVPR 2026] InfiniDepth: Arbitrary-Resolution and Fine-Grained Depth Estimation with Neural Implicit Fields

VINO-code. A Unified Visual Generator with Interleaved OmniModal Context

SpargeAttn. [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.

LongCat-Video. Python

Discrete-Diffusion-Forcing. Discrete Diffusion Forcing (D2F): dLLMs Can Do Faster-Than-AR Inference

Curator. Scalable data pre processing and curation toolkit for LLMs

LLaDA2.X. LLaDA2.0 is the diffusion language model series developed by InclusionAI team, Ant Group.

MatAnyone. [CVPR 2025] MatAnyone: Stable Video Matting with Consistent Memory Propagation

lmms-engine. A simple, unified multimodal models training engine. Lean, flexible, and built for hacking at scale.

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).

minimind-v. 👀「大模型」2小时从0训练65M参数的视觉多模态VLM！Train a 65M-parameter VLM from scratch in just 2h!

smollm. Everything about the SmolLM and SmolVLM family of models

torchtitan. A PyTorch native platform for training generative AI models

Pai-Megatron-Patch. The official repo of Pai-Megatron-Patch for LLM & VLM large scale training developed by Alibaba Cloud.

Emu3.5. Native Multimodal Models are World Learners

TAPIP3D. TAPIP3D: Tracking Any Point in Persistent 3D Geometry

MUG-V-Megatron-LM-Training. Official training code for MUG-V 10B video generation model. Built on Megatron-LM (v0.14.0) with production-ready distributed training for 10B DiT.

MUG-V. MUG-V 10B: High-efficiency Training Pipeline for Large Video Generation Models

Puffin. [ICLR 2026] Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation

DreamOmni2. This project is the official implementation of 'DreamOmni2: Multimodal Instruction-based Editing and Generation (CVPR2026 Highlight)''

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

nanochat. The best ChatGPT that $100 can buy.

simple_GRPO. A very simple GRPO implement for reproducing r1-like LLM thinking.

Bridge. Code of BRIDGE: Building Reinforcement-Learning Depth-to-Image Data Generation Engine for Monocular Depth Estimation

xtuner. A Next-Generation Training Engine Built for Ultra-Large MoE Models

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.

OmniWorld. [ICLR 2026] OmniWorld: A Multi-Domain and Multi-Modal Dataset for 4D World Modeling

minFM. HTML

WinT3R. Code of WinT3R: Window-Based Streaming Rrconstruction With Camera Token Pool

perception_models. State-of-the-art Image & Video CLIP, Multimodal Large Language Models, and More!

verifiers. Our library for RL environments + evals

Waver. Industry-level video foundation model for unified Text-to-Video (T2V) and Image-to-Video (I2V) generation.

prope. Cameras as Relative Positional Encoding

fairscale. PyTorch extensions for high performance and large scale training.

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

Wan2.2. Wan: Open and Advanced Large-Scale Video Generative Models

SpaTrackerV2. [ICCV 2025] SpatialTrackerV2: 3D Point Tracking Made Easy

SeC. [ICLR 2026] Official implementation of "SeC: Advancing Complex Video Object Segmentation via Progressive Concept Construction"

YUME. The official code of Yume

Muon. Muon is an optimizer for hidden layers in neural networks

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

CodeGeeX. CodeGeeX: An Open Multilingual Code Generation Model (KDD 2023)

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。

MediaCrawler. 小红书笔记 | 评论爬虫、抖音视频 | 评论爬虫、快手视频 | 评论爬虫、B 站视频 ｜ 评论爬虫、微博帖子 ｜ 评论爬虫、百度贴吧帖子 ｜ 百度贴吧评论回复爬虫 | 知乎问答文章｜评论爬虫

TradingAgents. TradingAgents: Multi-Agents LLM Financial Trading Framework

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

VQ-VLA. The offical repo for paper "VQ-VLA: Improving Vision-Language-Action Models via Scaling Vector-Quantized Action Tokenizers" (ICCV 2025)

WeClone. 🚀 One-stop solution for creating your AI twin from chat history 💡 Fine-tune LLMs with your chat logs to capture your unique style, then bind to a chatbot to bring your digital self to life.

Qwen-VL. The official repo of Qwen-VL (通义千问-VL) chat & pretrained large vision language model proposed by Alibaba Cloud.

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

picotron. Minimalistic 4D-parallelism distributed training framework for education purpose

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

EvoSearch-codes. An official implementation of EvoSearch: Scaling Image and Video Generation via Test-Time Evolutionary Search

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

DeepVerse. DeepVerse: 4D Autoregressive Video Generation as a World Model

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

BLIP3o. Official implementation of BLIP3o-Series

VACE. [ICCV 2025] Official implementations for paper: VACE: All-in-One Video Creation and Editing

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

nanoVLM. The simplest, fastest repository for training/finetuning small-sized VLMs.

diffusion-pipe. A pipeline parallel training script for diffusion models.

DiffSynth-Studio. Enjoy the magic of Diffusion models!

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

GRPO-Zero. Implementing DeepSeek R1's GRPO algorithm from scratch

aerial-megadepth. [CVPR 2025] AerialMegaDepth: Learning Aerial-Ground Reconstruction and View Synthesis

Pyramid-Flow. [ICLR 2025] Pyramidal Flow Matching for Efficient Video Generative Modeling

SegAnyMo. [CVPR 2025] Code for Segment Any Motion in Videos

Aether. [ICCV 2025 & ICCV 2025 RIWM Outstanding Paper] Aether: Geometric-Aware Unified World Modeling

Video-T1. [ICCV 2025] Video-T1: Test-Time Scaling for Video Generation

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

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

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

mega-sam. Code for the project "MegaSaM: Accurate, Fast and Robust Structure and Motion from Casual Dynamic Videos"

xAR. This repository includes the official implementation of our paper "Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation"

VisualThinker-R1-Zero. Explore the Multimodal “Aha Moment” on 2B Model

VADER. Video Diffusion Alignment via Reward Gradients. We improve a variety of video diffusion models such as VideoCrafter, OpenSora, ModelScope and StableVideoDiffusion by finetuning them using various reward models such as HPS, PickScore, VideoMAE, VJEPA, YOLO, Aesthetics etc.

VLM-R1. Solve Visual Understanding with Reinforced VLMs

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!

ktransformers. A Flexible Framework for Experiencing Heterogeneous LLM Inference/Fine-tune Optimizations

unsloth. Unsloth Studio is a web UI for training and running open models like Gemma 4, Qwen3.6, DeepSeek, gpt-oss locally.

R1-V. Witness the aha moment of VLM with less than $3.

CUT3R. Official implementation of Continuous 3D Perception Model with Persistent State

ReasonFlux. [NeurIPS 2025 Spotlight] LLM post-training suite — featuring ReasonFlux, ReasonFlux-PRM, and ReasonFlux-Coder.

LIMO. [COLM 2025] LIMO: Less is More for Reasoning

open-r1. Fully open reproduction of DeepSeek-R1

s1. s1: Simple test-time scaling

TinyZero. Minimal reproduction of DeepSeek R1-Zero

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.

MVision. 机器人视觉 移动机器人 VS-SLAM ORB-SLAM2 深度学习目标检测 yolov3 行为检测 opencv PCL 机器学习 无人驾驶

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.

genesis-world. Simulation platform for general-purpose robotics & embodied AI learning.

lerobot. 🤗 LeRobot: Making AI for Robotics more accessible with end-to-end learning

Janus. Janus-Series: Unified Multimodal Understanding and Generation Models

GameGen-X.

MoGe. [CVPR'25 Oral] MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision

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

ai-for-grant-writing. A curated list of resources for using LLMs to develop more competitive grant applications.

Sana. SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

mochi. The best OSS video generation models, created by Genmo

modded-nanogpt. NanoGPT (124M) in 90 seconds

RoboticsDiffusionTransformer. RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation