Projects

### Overview: This repository contains the official implementation of the "3D Gaussian Splatting for Real-Time Radiance Field Rendering" paper, authored by Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, and George Drettakis. The project is part of the FUNGRAPH research initiative and aims to achieve state-of-the-art visual quality while maintaining real-time rendering capabilities at 1080p resolution. ### Key features: - **3D Gaussian Splatting**: Represents the scene with 3D Gaussians to preserve continuous volumetric radiance fields while avoiding unnecessary computation in empty space. - **Interleaved Optimization**: Optimizes the 3D Gaussians to achieve an accurate representation of the scene. - **Fast Visibility-Aware Rendering**: Supports anisotropic splatting and accelerates training and real-time rendering. ### Tech stack: Python ### Why it matters: This implementation is crucial for achieving high-quality real-time rendering of novel-view scenes, particularly for applications requiring fast and efficient rendering at high resolutions.

### Overview: Orchestrate Claude, Codex, and Grok in one app, providing a unified interface for accessing multiple AI models. Happy Harness is open-source and reuses existing subscriptions, making it a cost-effective solution for developers. ### Key Features: - **Multi-provider support**: Access Astra, Fable, and Grok models in one app, each with its own prompts and tools. - **Subscription reuse**: Sign in with existing Claude, Codex, and Grok subscriptions to avoid additional bills. - **Open-source MIT license**: Runs on your own hardware and keeps your projects simple and ordinary. - **End-to-end encrypted mobile app**: Sessions are stored on your device, and communication between devices is encrypted. - **Natively multiplayer**: Invite colleagues or friends to join sessions, and both can interact with the same agent. ### Tech Stack: TypeScript ### Why it Matters: Happy Harness simplifies the process of integrating multiple AI models into a single application, making it easier for developers to leverage the capabilities of Claude, Codex, and Grok without the complexity of managing multiple subscriptions or separate applications.

Overview: PaddlePaddle is an industrial platform for deep learning and machine learning, designed for both single-machine and distributed training, as well as cross-platform deployment. Key features: - Unified Dynamic/Static Graphs and Automatic Parallelism - Integrated Training and Inference for Large Models - High-Order Differentiation for Scientific Computing - Neural Network Compiler - Heterogeneous Multi-Chip Adaptation Tech stack: C++ Why it matters: PaddlePaddle is a powerful tool for developing and deploying deep learning models, with features that are highly relevant to industrial applications and have been widely adopted by various sectors.

### Overview: PyG is a comprehensive library built on PyTorch for deep learning on graphs and other irregular structures, also known as geometric deep learning. It provides a unified API for easy and efficient training of Graph Neural Networks (GNNs) on various applications related to structured data. ### Key Features: - **Easy-to-use and unified API**: All it takes is 10-20 lines of code to get started with training a GNN model. - **Comprehensive and well-maintained GNN models**: Most of the state-of-the-art GNN architectures are implemented, ready for application. - **Great flexibility**: Existing models can be easily extended, and modifications or creation of new architectures are simple due to its message passing API and utility functions. - **Large-scale real-world GNN models**: Supports learning on diverse types of graphs, including scalable GNNs for graphs with millions of nodes, dynamic GNNs for node predictions over time, and heterogeneous GNNs with multiple node types and edge types. ### Tech Stack: PyTorch ### Why It Matters: Whether you are a machine learning researcher or a first-time user of machine learning toolkits, PyG is a valuable resource for machine learning on graph-structured data. It simplifies the process of training GNN models, making it accessible to both experienced practitioners and beginners. The library's unified API and comprehensive models make it a versatile tool for various applications in the field of geometric deep learning.

### Overview: 本项目是一份全面的深度学习笔记，涵盖了CV、NLP、大模型和Agent等领域的知识。笔记视频由土堆、李沐、吴恩达和大飞主讲，内容详尽，适合学习和实践。 ### Key features: - **深度学习视频**：包含Pytorch、深度学习和大模型Agent的详细讲解。 - **数据集**：无偿分享的深度学习相关数据集，包括图像、文本和模型训练数据。 - **目录插件**：快速跳转到相应章节，方便学习。 ### Tech stack: Jupyter Notebook - 使用Anaconda的Jupyter Notebook进行笔记的编写和展示。 ### Why it matters: 这份笔记是深度学习领域的全面指南，适合初学者和进阶学习者。通过视频讲解和详细笔记，能够快速掌握深度学习的核心知识和实践技能。 ### Repository: AccumulateMore/CV - GitHub Stars: 24,045 - GitHub Description: 超级全面的深度学习笔记【土堆 Pytorch】【李沐 动手学深度学习】【吴恩达 深度学习】【大飞 大模型Agent】 ### Notes: - **图片显示问题**：在线观看时可能会出现图片显示不完整的问题，建议下载笔记到本地查看。 - **Jupyter Notebook问题**：在使用Pycharm打开笔记时，图片不会正常显示，需要使用Anaconda的Jupyter Notebook打开。 - **目录插件**：安装目录插件后

### Overview: CosyVoice is a state-of-the-art multi-lingual large voice generation model that supports zero-shot multilingual speech synthesis. It offers advanced features such as language coverage, content consistency, naturalness, pronunciation inpainting, text normalization, and bi-streaming capabilities. ### Key Features: - **Language Coverage**: Covers 9 common languages (Chinese, English, Japanese, Korean, German, Spanish, French, Italian, Russian), 18+ Chinese dialects/accents. - **Content Consistency & Naturalness**: Achieves state-of-the-art performance in content consistency, speaker similarity, and prosody naturalness. - **Pronunciation Inpainting**: Supports pronunciation inpainting of Chinese Pinyin and English CMU phonemes, providing more controllability. - **Text Normalization**: Supports reading of numbers, special symbols, and various text formats without a traditional frontend module. - **Bi-Streaming**: Supports both text-in streaming and audio-out streaming, with latency as low as 150ms. - **Instruct Support**: Supports various instructions such as languages, dialects, emotions, speed, volume, etc. ### Tech Stack: Python ### Why It Matters: CosyVoice is a groundbreaking model that significantly improves the quality and consistency of speech synthesis, making it ideal for applications requiring high-fidelity audio output in real-world scenarios.

Overview: ncnn is a high-performance neural network inference framework optimized for mobile, embedded, and desktop deployment. It is used in many Tencent applications. Key features: - Supports PyTorch and ONNX models - Runs on CPU and Vulkan GPU backends - Provides tools like pnnx for model conversion - Efficiently deploys deep learning models on various devices Tech stack: C++ Why it matters: ncnn is crucial for efficient deployment of deep learning models across various devices, making it essential for mobile, embedded, and desktop applications.

### Overview: This curated list ranks 920 open-source projects across various machine learning categories, including frameworks, data visualization, text data, image data, and more. Updated weekly, it provides a comprehensive overview of the best tools and libraries for machine learning enthusiasts. ### Key features: - Machine Learning Frameworks - Data Visualization - Text Data & NLP - Image Data - Graph Data - Audio Data - Geospatial Data - Financial Data - Time Series Data - Medical Data - Tabular Data - Optical Character Recognition - Data Containers & Structures - Data Loading & Extraction - Web Scraping & Crawling - Data Pipelines & Streaming - Distributed Machine Learning - Hyperparameter Optimization & AutoML - Reinforcement Learning - Recommender Systems - Privacy Machine Learning - Workflow & Experiment Tracking - Model Serialization & Deployment - Model Interpretability - Vector Similarity Search (ANN) - Probabilistics & Statistics - Adversarial Robustness - GPU & Accelerator Utilities - TensorFlow Utilities ### Tech stack: Python ### Why it matters: This curated list provides a valuable resource for developers looking to explore and utilize the best machine learning tools available. It helps in quickly identifying and integrating the most suitable libraries and frameworks into their projects, ensuring efficiency and effectiveness in machine learning development.

### Overview: Vanna 2.0 is a powerful tool that turns natural language queries into SQL code and data insights, making it easier to extract meaningful information from your database. ### Key Features: 1. **User-Aware Permissions**: Queries automatically filter based on user permissions. 2. **Modern Web Interface**: Beautiful pre-built `<vanna-chat>` component. 3. **Real-Time Streaming**: Tables, charts, and progress updates streamed in real-time. 4. **Enterprise Security**: Row-level security, audit logs, and rate limiting. 5. **Production-Ready**: FastAPI integration, observability, and lifecycle hooks. 6. **Custom Tools**: Extend the `Tool` base class and use lifecycle hooks for custom functionality. 7. **LLM Integration**: Works with any LLM, including OpenAI, Anthropic, and others. 8. **Database Support**: Supports PostgreSQL, MySQL, Snowflake, BigQuery, Redshift, SQLite, Oracle, SQL Server, DuckDB, ClickHouse, and more. 9. **Framework-agnostic**: Compatible with React, Vue, and plain HTML. 10. **Custom Agent**: Allows you to define your own agent with your database. ### Why It Matters: - **Instant Start**: Production chat interface, custom agent, and embeddable in any webpage. - **Security**: User-aware at every layer, row-level security, audit logs, and rate limiting. - **UI**: Pre-built `<vanna-chat>` component, streaming tables and charts, responsive and customizable. - **Stack Compatibility**: Works with any LLM

### Overview: 《明日方舟》小助手，全日常一键长草！| A one-click tool for the daily tasks of Arknights, supporting all clients. ### Key features: - 一键完成全部日常任务 - 智能基建换班，自动计算干员效率 - 自动公招，支持加急许可 - 支持手动识别公招界面和干员列表 - 支持识别养成材料并导出 - 访问好友、收取信用及购物、领取日常奖励 - 肉鸽全自动刷源石锭和等级，自动烧水和凹直升 - 选择作业 JSON 文件，自动抄作业 - 支持多种接口调用，方便集成 ### Tech stack: C++ ### Why it matters: 这款工具简化了《明日方舟》的日常操作，提高了效率，是玩家和开发者日常管理的重要工具。

### Overview: DeepFace is a lightweight Python library for face recognition and facial attribute analysis, including age, gender, emotion, and race detection. It utilizes state-of-the-art models such as VGG-Face, FaceNet, OpenFace, DeepFace, DeepID, ArcFace, Dlib, and SFace, among others. ### Key Features: - **Face Recognition**: Detects and classifies faces with high accuracy. - **Facial Attributes**: Analyzes age, gender, emotion, and race from images. - **State-of-the-Art Models**: Integrates advanced models like VGG-Face, FaceNet, and DeepFace. - **Ease of Use**: Simple and intuitive API for developers. ### Tech Stack: - Python ### Why It Matters: DeepFace is a comprehensive solution for developers looking to integrate face recognition and attribute analysis into their projects, making it easier to implement advanced facial recognition and attribute detection in Python applications.

### Overview: Marimo is a modern, AI-native Python notebook that allows you to run reproducible experiments, query data, execute as scripts, and deploy as apps. It is designed to be git-friendly, storing code and outputs as pure Python, making it easy to version and share. ### Key Features: - **Reactive Execution**: Run a cell, and marimo automatically runs dependent cells or marks them as stale. - **Interactive UI**: Bind sliders, tables, plots, and more to Python code. - **Git-Friendly Storage**: Stored as pure Python, making it easy to version and share. - **Data Querying**: Query dataframes, databases, warehouses, or lakehouses with SQL. - **AI Pairing**: Pair with AI agents like Claude Code or use built-in AI features. - **Reproducible Execution**: No hidden state, deterministic execution, and built-in package management. - **Executable as Script**: Execute as a Python script with CLI arguments. - **Shareable Apps**: Deploy as an interactive web app or slides, run in the browser via WASM. - **Reusable Functions**: Import functions and classes from one notebook to another. ### Tech Stack: Python ### Why It Matters: Marimo is a powerful tool for data scientists, researchers, and developers who need a modern, git-friendly, and AI-native environment for their work. Its ability to run experiments, query data, and deploy as apps makes it a versatile solution for various projects.

Overview: This repository, hosted on GitHub, contains code for Computer Vision, Deep learning, and AI research articles shared on the LearnOpenCV blog. It is maintained by BigVision.AI, a computer vision and AI consulting company that specializes in designing, building, optimizing, and deploying production-ready AI solutions. Key features: - OpenCV code examples - C++ and Python tutorials - Real-world projects and solutions Tech stack: C++ and Python Why it matters: This repository provides valuable resources for developers looking to enhance their computer vision and AI skills through practical examples and real-world applications.

### Overview: Hypit is a platform that allows AI agents like Claude Code and Codex to create viral videos by cloning any existing video. It provides a complete workflow from scratch, including footage, captions, B-roll, and effects, all tied to the original video's words. ### Key Features: - **Cloning Videos**: Quickly generate 100 variants of a video by swapping faces, words, and B-roll. - **Command Line Interface**: A single command generates 100 variants in one step. - **100M Views**: Achieve 100 million views with just one command. ### Tech Stack: TypeScript ### Why It Matters: Hypit simplifies the process of creating viral videos, making it accessible to anyone with a video and an AI agent. The platform's ability to generate 100 variants in one command and achieve 100 million views showcases its effectiveness and ease of use. ### Repository: The repository for Hypit is hosted on GitHub, with 23,092 stars. It provides a demo, quickstart guide, and detailed development documentation. ### GitHub Stars: - **Stars**: 23,092 - **Description**: Clone any viral video with AI agents. Not just a script, the whole workflow: swap the face, the words, the B-roll, ship 100 variants in one command, and get your 100M views. ### Primary Language: TypeScript ### GitHub Description: Clone any viral video with AI agents. Not just

Overview: Go Micro is a Go framework designed for the development of distributed systems, offering RPC, service discovery, messaging, storage, and configuration features. Key features: - Services: RPC with streaming, message encoding, and service discovery. - Agents: models, conversation history, tools, and agent execution. - Workflows: ordered steps, retries, and checkpoints. Tech stack: Go Why it matters: Go Micro simplifies the development of distributed systems by providing a comprehensive set of tools and features, making it easier to build, manage, and scale microservices. Repository: micro/go-micro Primary language: Go GitHub stars: 23086 GitHub description: A framework for agents, tools and services README excerpt:

### Overview: MaxKB is an open-source platform designed for building enterprise-grade agents. It integrates Retrieval-Augmented Generation (RAG) pipelines, robust workflows, and advanced MCP tool-use capabilities, making it suitable for intelligent customer service, corporate knowledge bases, academic research, and education. ### Key Features: - **RAG Pipeline**: Supports direct document uploads and automatic crawling of online documents, with text splitting and vectorization features to reduce hallucinations. - **Agentic Workflow**: Equipped with a powerful workflow engine, function library, and MCP tool-use capabilities, enabling the orchestration of AI processes for complex business scenarios. - **Seamless Integration**: Facilitates zero-code rapid integration into third-party business systems, enhancing existing systems with intelligent Q&A capabilities. - **Model-Agnostic**: Supports various large models, including private models and public models. - **Multi Modal**: Native support for text, image, audio, and video inputs and outputs. ### Tech Stack: - Frontend: Vue.js - Backend: Python / Django - LLM Framework: LangChain ### Why It Matters: MaxKB is widely applied in various industries, including intelligent customer service, corporate knowledge bases, academic research, and education, providing a superior smart Q&A interaction experience and enhancing user satisfaction through seamless integration and model-agnostic support. ### GitHub Stars: 22,989 ### Repository: 1Panel-dev/MaxKB ### Primary Language: Python ### GitHub Description: 🔥 MaxKB is an open-source platform for building enterprise-grade agents. It integrates Retrieval

### Overview: This repository tracks the progress in Natural Language Processing (NLP), providing datasets and the current state-of-the-art for common NLP tasks. It serves as a comprehensive resource for researchers and practitioners in the field. ### Key features: - Automatic Speech Recognition - Constituency Parsing - Coreference Resolution - Data-to-Text Generation - Dependency Parsing - Domain Adaptation - Entity Linking - Grammatical Error Correction - Information Extraction - Intent Detection and Slot Filling - Keyphrase Extraction and Generation - Language Modeling - Lexical Normalization - Machine Translation - Missing Elements - Multi-Task Learning - Multi-Modal - Named Entity Recognition - Natural Language Inference - Part-of-Speech Tagging - Paraphrase Generation - Question Answering - Relation Prediction - Relationship Extraction - Semantic Textual Similarity - Semantic Parsing - Semantic Role Labeling - Sentiment Analysis - Shallow Syntax - Simplification - Stance Detection - Summarization - Taxonomy Learning - Temporal Processing - Text Classification - Word Sense Disambiguation ### Tech stack: Python ### Why it matters: This repository is essential for anyone involved in NLP research and development, providing a centralized platform for tracking progress and sharing results across various NLP tasks and datasets.

### Overview: This repository contains the code for the paper "Dropping Anchor and Spherical Harmonics for Sparse-view Gaussian Splatting" by [Author(s)]. The code is designed to implement the Gaussian Splatting method for sparse-view reconstruction, focusing on improving the accuracy and efficiency of the process. ### Key Features: - **Sparse-view Gaussian Splatting Implementation**: The repository provides the complete implementation of the Gaussian Splatting method for sparse-view reconstruction. - **Anchor Dropping and Spherical Harmonics**: The code includes the techniques of dropping anchors and using spherical harmonics to enhance the accuracy of the sparse-view reconstruction. ### Tech Stack: - Python - NumPy - OpenCV ### Why it Matters: This project is crucial for researchers and developers working on sparse-view reconstruction, as it offers a more accurate and efficient method compared to traditional approaches. The techniques used in this repository can be applied to various applications in computer vision, such as 3D reconstruction from low-resolution images or videos. ### Repository: Repository: [amusi/CVPR2026-Papers-with-Code](https://github.com/amusi/CVPR2026-Papers-with-Code) GitHub Stars: 22,903

### Overview: 本项目旨在整理和梳理中文大语言模型（LLM）相关的开源资源，涵盖模型、应用、数据集和教程等。目前收录的资源已超过100个，覆盖了多种类型的模型和应用，包括文本LLM、多模态LLM、垂直领域微调、LangChain应用、数据集和评测等。 ### Key features: - 文本LLM模型 - ChatGLM - ChatGLM2-6B - 多模态LLM模型 - LLaMA - Baichuan - Qwen - BLOOM - Aquila - InternLM - Mixtral - Yi - DeepSeek - XVERSE - MiniMax - 应用 - 垂直领域微调 - 医疗 - 法律 - 金融 - 教育 - 科技 - 电商 - 网络安全 - 农业 - LangChain应用 - 其他应用 - 数据集 - 预训练数据集 - SFT数据集 - 偏好数据集 - LLM训练微调框架 - LLM推理部署框架 - LLM评测 - LLM教程 - LLM基础知识 - 提示工程教程 - LLM应用教程

### Overview: NVIDIA NemoClaw provides a secure and managed environment for running AI agents inside NVIDIA OpenShell, making it easier and safer for developers. ### Key Features: - **NVIDIA OpenShell Integration**: Securely runs supported AI agents inside the OpenShell sandbox. - **Managed Inference**: Managed inference ensures stable and consistent performance. - **Network Policy**: Controls network traffic to and from the agent. - **Managed Integrations**: Simplifies integration with external systems. - **State Persistence**: Securely stores agent state. - **Lifecycle Operations**: Automates agent lifecycle management. ### Tech Stack: TypeScript ### Why it Matters: NVIDIA NemoClaw simplifies the process of running AI agents securely inside NVIDIA OpenShell, making it easier for developers to focus on their code without worrying about the complexities of sandbox management and security.