Projects

Overview: Karate Club is an open-source Python framework designed for unsupervised learning on graphs, focusing on community detection and node embeddings. It aims to provide a comprehensive solution for small-scale graph mining research across various fields such as network science, data mining, artificial intelligence, and machine learning. Key features: - Community detection methods - Node embeddings - Integration with NetworkX for easy network manipulation - Support for both overlapping and non-overlapping community detection Tech stack: Python Why it matters: Karate Club offers a powerful toolset for researchers and practitioners looking to perform unsupervised learning on graph data, making it an essential resource for advancing graph mining and related fields. Repository: benedekrozemberczki/karateclub Primary language: Python GitHub stars: 2291 GitHub description: Karate Club: An API Oriented Open-source Python Framework for Unsupervised Learning on Graphs (CIKM 2020) README excerpt:   [](https://github.com/benedekrozemberczki/karateclub/archive/master.zip) [![Arxiv](https://img.shields.io/badge/ArXiv-2003.048

### Overview: MMAudio is a cutting-edge model that generates synchronized audio from video and/or text inputs, achieving high-quality synthesis. ### Key Features: - **Multimodal Joint Training**: Trains on a wide range of audio-visual and audio-text datasets. - **Synchronization Module**: Aligns generated audio with video frames to ensure synchronization. - **High-Quality Synthesis**: Produces high-quality audio that matches the visual content, making it perfect for applications like voice-to-video synthesis. ### Tech Stack: Python ### Why It Matters: MMAudio is a significant advancement in the field of multimodal audio synthesis, offering a solution that can be applied to various applications, including voice-to-video synthesis, enhancing user experience in video games, and more. Its ability to generate synchronized audio from video and text inputs sets it apart from existing solutions, making it a valuable tool for developers and researchers looking to improve the quality of video-to-audio synthesis.

### Overview: This repository contains a collection of World Models used for Autonomous Driving and related applications. It serves as a comprehensive resource for researchers, providing a platform to track and benchmark various methods. The repository includes papers from the arXiv and other sources, with a focus on recent advancements in the field. ### Key Features: - **World Models for Autonomous Driving**: A repository for recording, tracking, and benchmarking World Models used in Autonomous Driving and related applications. - **Contributions Welcome**: Contributions in any form are welcome to make the repository more comprehensive and useful. - **Citation**: If you find this repository useful, please consider giving a star ⭐ and a citation. ### Tech Stack: - **GitHub**: Used for hosting and managing the repository. ### Why It Matters: This repository is crucial for researchers and developers working on Autonomous Driving and related technologies. It provides a centralized location for tracking and benchmarking the latest methods, ensuring that the community remains informed about the latest advancements in the field. ### Repository: - **Repository Name**: LMD0311/Awesome-World-Model - **GitHub Stars**: 2288 - **GitHub Description**: Collect some World Models for Autonomous Driving (and Robotic, etc.) papers. - **Primary Language**: Unknown - **GitHub URL**: [https://github.com/LMD0311/Awesome-World-Model](https://github.com/LMD0311/Awesome-World-Model)

Overview: This repository is a curated list of pretrained sentence and word embedding models, aiming to provide a comprehensive and up-to-date resource for developers. Key features: - Comprehensive list of pretrained models - Table of contents for easy navigation - Information about general framework, word embeddings, OOV handling, context-aware embeddings, pooling methods, encoders, evaluation, and vector mapping Tech stack: Python Why it matters: This list serves as a valuable resource for developers looking to incorporate sentence and word embedding models into their projects, offering a curated and up-to-date collection of models and their training codes. Repository: Separius/awesome-sentence-embedding GitHub stars: 2288

Overview: Lagent is a lightweight framework designed for building LLM-based agents, inspired by the PyTorch design philosophy. It simplifies the workflow by focusing on creating layers and defining message passing between them in a Pythonic way. Key features: - Lightweight framework for LLM-based agent development - Uses `AgentMessage` for communication between agents - Supports custom message aggregation with `DefaultAggregator` - Memory management in `Agent` class - Integration with PyTorch for seamless model training and inference Tech stack: Python Why it matters: Lagent simplifies the process of creating LLM-based agents, making it easier for developers to focus on the core functionality without getting bogged down in complex details. The integration with PyTorch allows for seamless model training and inference, making it suitable for a wide range of applications. Repository: InternLM/lagent GitHub stars: 2286 GitHub description: A lightweight framework for building LLM-based agents

Overview: COCO Annotator is a versatile web-based image annotation tool designed for efficient image segmentation and object detection. It supports the creation of training datasets in the COCO format, making it ideal for use in computer vision projects. Key features: - Directly export to COCO format - Segmentation of objects - Ability to add key points - Useful API endpoints for data analysis - Import datasets already annotated in COCO format - Annotate image segments with multiple labels simultaneously - Track object instances - Label disconnected objects as a single instance Tech stack: Vue Why it matters: COCO Annotator simplifies the process of creating and managing image annotations, making it an essential tool for developers and researchers in the field of computer vision. Its ability to efficiently label images and generate well-structured datasets is highly valuable, particularly for training machine learning models.

### Overview: `dstack` is a unified orchestration layer for GPU provisioning and orchestration that works with any GPU cloud, Kubernetes, or on-prem clusters. It streamlines development, training, and inference, and is compatible with any hardware, open-source tools, and frameworks. ### Key features: - GPU provisioning and orchestration across different cloud providers and on-prem clusters. - Support for NVIDIA, AMD, Google TPU, and Tenstorrent accelerators. - Compatibility with any hardware, open-source tools, and frameworks. - Streamlined development, training, and inference process. - Configurable backends for GPU clouds and Kubernetes clusters. - SSH fleets for on-prem servers without backend configuration. ### Tech stack: Python ### Why it matters: `dstack` simplifies the management and orchestration of GPU resources, making it easier to deploy and scale machine learning models across various environments. Its compatibility with diverse hardware and frameworks ensures broad applicability, while its modular architecture allows for easy integration with existing infrastructure.

Overview: 本项目旨在通过多种模型（HMM, CRF, BiLSTM, BiLSTM+CRF）解决中文命名实体识别问题，包括简历数据集的收集、模型训练与评估、以及预测结果的展示。 Key features: - 支持多种模型实现（HMM, CRF, BiLSTM, BiLSTM+CRF） - 数据集包含简历数据，格式为字及其标注 - 模型训练与评估结果详细展示，包括准确率、召回率、F1分数和混淆矩阵 - 可以通过Ensemble方法将多个模型的预测结果结合，使用投票表决法得出最终预测 Tech stack: Python Why it matters: 通过多种模型的结合，本项目在中文命名实体识别领域提供了全面的解决方案，适用于多种应用场景，特别是对于简历数据集的处理和预测结果的准确性和效率具有显著提升。 Repository: luopeixiang/named_entity_recognition

### Overview: 深度有趣 - 人工智能实战项目合集，涵盖182页的课程说明文档，包括18个实战项目，使用Python3、TensorFlow1.9和Keras2.2。项目包括数据和代码，分为完整版和精简版两个版本，提供知乎专栏和付费视频学习资源。 ### Key features: - 实战项目合集 - 项目详细说明 - 开源项目资源 - 项目数据和代码 - 项目目录详细 - 项目实战经验 ### Tech stack: Python3, TensorFlow1.9, Keras2.2 ### Why it matters: 通过深度有趣，你可以学习到人工智能领域的实战项目，掌握Python、TensorFlow和Keras的实战技能，提升编程能力，实现深度学习项目。

Overview: BGSLibrary is a C++ background subtraction library that provides easy-to-use and extensible tools for detecting moving objects in video streams. It includes wrappers for Python, MATLAB, Java, and a Qt GUI, making it compatible with OpenCV versions 2.4.x, 3.x, and 4.x. Key features: - Easy-to-use C++ framework - Extensible platform for background subtraction techniques - Compatible with OpenCV versions 2.4.x, 3.x, and 4.x - Wrappers for Python, MATLAB, Java, and Qt GUI - Compatible with Windows, Linux, and Mac OS X systems Tech stack: C++ Why it matters: BGSLibrary simplifies the process of implementing background subtraction techniques in computer vision applications, making it accessible to researchers and developers without requiring deep knowledge of OpenCV or background subtraction algorithms.

Overview: ECCV 2026 论文和开源项目合集，包含来自 ECCV 2024、ECCV 2022 和 ECCV 2020 的论文和代码。欢迎提交issue分享 ECCV 2024 论文和开源项目。 Key features: - ECCV 2024 论文和代码合集 - ECCV 2022 论文和代码合集 - ECCV 2020 论文和代码合集 Tech stack: Python, PyTorch, OpenCV, TensorFlow,以及其他相关技术栈 Why it matters: ECCV 2026 论文和开源项目合集为计算机视觉领域的研究者和开发者提供了丰富的资源，有助于推动技术的发展和应用。

### Overview: EdgeQuake is a high-performance GraphRAG inspired from LightRag written in Rust. It transforms documents into intelligent knowledge graphs, enabling superior retrieval and generation. Designed for superior performance and ease of use, EdgeQuake integrates seamlessly with any LLM, making it a versatile tool for knowledge discovery and retrieval. ### Key Features: - **GraphRAG Inspired from LightRag**: Leveraging advanced graph algorithms for efficient knowledge graph construction. - **Rust Implementation**: Built with Rust, ensuring performance and reliability. - **Document-to-Graph Transformation**: Quickly converting documents into knowledge graphs. - **Superior Retrieval and Generation**: Seamlessly integrating with any LLM for enhanced results. ### Tech Stack: - Rust ### Why It Matters: EdgeQuake empowers developers and researchers to quickly transform their documents into intelligent knowledge graphs, enhancing their ability to retrieve and generate insights. Its performance and ease of integration make it a valuable tool for anyone looking to unlock the power of knowledge graphs. Repository: raphaelmansuy/edgequake Primary language: Rust GitHub stars: 2279 GitHub description: EdegQuake 🌋 High-performance GraphRAG inspired from LightRag written in Rust; Transform documents into intelligent knowledge graphs for superior retrieval and generation.

Overview: TransmogrifAI is an AutoML library for building modular, reusable, strongly typed machine learning workflows on Apache Spark, developed with a focus on accelerating machine learning developer productivity through machine learning automation and an API that enforces compile-time type-safety, modularity, and reuse. Key features: - AutoML library for building machine learning workflows - Supports Apache Spark - Enforces compile-time type-safety - Modularity and reuse of code - Accurate predictions with almost 100x reduction in time Tech stack: Scala Why it matters: TransmogrifAI is designed to help developers build production-ready machine learning applications quickly, without the need for extensive machine learning expertise. It automates the process of creating machine learning models, ensuring that the models are accurate and efficient, while also reducing the time required to build and deploy the models. GitHub stars: 2280 Repository: salesforce/TransmogrifAI

### Overview: Streamlit is a Python-based framework for building interactive web applications. It allows developers to create data-driven applications without the need for complex backend development, making it an excellent tool for technical writers and data scientists alike. ### Key Features: - **Interactive Web Applications**: Streamlit generates web applications that can be deployed directly to the cloud. - **Python Backend**: Built on Python, it integrates seamlessly with other Python libraries. - **Easy Deployment**: Streamlit apps can be deployed to platforms like Glitch, AWS, or Google Cloud. - **Automated CI/CD**: Streamlit supports CI/CD pipelines for continuous integration and deployment. - **Docker Container**: Streamlit apps can be run in Docker containers for easy deployment and isolation. ### Tech Stack: Python ### Why it Matters: Streamlit simplifies the process of building data-driven applications, making it an invaluable tool for technical writers and data scientists. Its ease of use and integration with Python make it an excellent choice for creating dynamic and interactive web applications without the need for complex backend development.

### Overview: This project is a collection of papers and code related to wireless communication and deep learning. It aims to provide a comprehensive overview of recent research in the field, focusing on applications of deep learning in wireless communication. ### Key Features: - **Code Repositories**: Contains open-source code for various papers on wireless communication and deep learning. - **Paper Categories**: Organized into different categories such as Physical Layer Optimization, Resource and Network Optimization, Distributed Learning Algorithms, Multiple Access Scheduling and Routing, Machine Learning for Network Slicing, Software-Defined Networking, Emerging Communication Systems, and Secure Machine Learning. - **Community Contributions**: Encourages contributions from the community by welcoming suggestions for additional papers and code. ### Tech Stack: The project uses various programming languages and tools, including Python, Git, and GitHub Actions for automatic code updates. ### Why It Matters: - **Access to Research**: Provides access to cutting-edge research papers and code, allowing researchers to stay updated and contribute to the field. - **Ease of Use**: Facilitates easy access to the latest research in wireless communication and deep learning, making it easier for students and researchers to learn and apply these techniques. - **Community Involvement**: Encourages community participation by welcoming contributions from the community, fostering a collaborative environment. ### Repository: The repository is maintained on GitHub and has received 2,275 stars, indicating significant user interest and engagement. ### TODO: - **Paper Classification**: Continue to organize papers into different categories. - **Code Repositories**: Add more relevant papers and code repositories. - **Additional

Overview: Dash is a self-learning data agent built with systems engineering principles, providing context-rich answers grounded in six layers of context and improving with every query. It grounds answers in 6 layers of context and improves with every query, making it a valuable tool for developers and data enthusiasts. Key features: - Context-rich answers grounded in six layers of context - Self-learning loop that improves with every query - Built with systems engineering principles Tech stack: Python Why it matters: Dash addresses the limitations of raw LLMs by providing context-rich answers and a self-learning loop that improves with every query, making it a valuable tool for developers and data enthusiasts.

Overview: RePaint is an official PyTorch implementation of the "RePaint: Inpainting using Denoising Diffusion Probabilistic Models" paper presented at CVPR 2022. This project aims to fill in missing parts of images using diffusion models. Key features: - Inpainting functionality using Denoising Diffusion Probabilistic Models - Supports ImageNet, CelebA-HQ, and Places2 datasets with various mask configurations - Ready-to-use config files for ImageNet, CelebA-HQ, and Places2 - Configurable noise levels and mask types Tech stack: Python, PyTorch, GANs, and diffusion models Why it matters: RePaint provides a practical solution for image inpainting, demonstrating the potential of diffusion models in computer vision tasks.

Overview: The Simd library is a comprehensive image processing and machine learning toolkit written in C++ that leverages the power of SIMD (Single Instruction, Multiple Data) extensions for enhanced performance. Key features: - Supports SSE, AVX, AVX-512, AMX for x86/x64 - NEON, SVE for ARM - SVE2 for ARM - HVX for Hexagon - C++ API and C++ classes for easy access - Dynamic and static linking - Windows and Linux support - MSVS, G++, and Clang compilers - CMake build systems Tech stack: C++ Why it matters: The Simd library offers optimized algorithms for image processing and machine learning tasks, significantly improving performance through the use of SIMD extensions. This makes it particularly valuable for developers working on high-performance applications in fields such as computer vision, image processing, and machine learning.

Overview: 《自然语言处理入门》详细笔记，书中通过白话阐述算法模型，从中文分词、词性标注、命名实体识别、信息抽取、文本聚类、文本分类、句法分析等热门问题入手，帮助读者快速掌握NLP专业知识。 Key features: - 中文分词 - 词性标注 - 命名实体识别 - 信息抽取 - 文本聚类 - 文本分类 - 句法分析 Tech stack: Python Why it matters: 本书是业界良心之作，内容通俗易懂，适合初学者快速掌握NLP知识，提高实际工作中应用NLP的能力。

Overview: NanoNeuron is a simplified version of the Neuron concept, designed to demonstrate machine learning principles using JavaScript. It converts temperature values from Celsius to Fahrenheit through 7 simple functions. Key features: - 7 simple JavaScript functions for model prediction, cost calculation, forward/backwards propagation, and training - Pure and simple JavaScript implementation without 3rd-party libraries or external dependencies - No complete guide to machine learning, focusing on basic understanding and recognizing the math behind machine learning Tech stack: JavaScript Why it matters: This project provides a basic understanding of machine learning concepts through a simplified implementation of the Neuron model, making it accessible for developers to learn and recognize the underlying math behind machine learning. Repository: trekhleb/nano-neuron GitHub stars: 2273