Projects

### Overview: Machine Learning Open Source University is an educational resource for enthusiasts of machine learning, offering free access to a comprehensive curriculum covering various aspects of the field. ### Key Features: - **Mathematics**: Provides foundational knowledge in statistics and linear algebra. - **Machine Learning**: Includes courses on deep learning, natural language processing, and large language models. - **Deep Learning**: Offers advanced topics in neural networks and generative AI. - **Natural Language Processing**: Focuses on understanding and generating human language. - **Reinforcement Learning**: Explains algorithms for decision-making in uncertain environments. - **LLM (Large Language Model)**: Covers advanced topics in language understanding and generation. - **Generative and Agentic AI**: Discusses AI systems that can generate and interact with the world. - **Books**: Lists recommended books for deepening understanding in specific areas. - **ML in Production**: Guides practitioners on integrating machine learning into real-world applications. - **Quantum ML**: Explores the intersection of quantum computing and machine learning. - **Datasets**: Provides a collection of datasets for practical applications. - **Other Useful Websites**: Links to additional resources and forums. - **Other Useful Git Repositories**: Suggests repositories for further learning. - **Blogs and Webinars**: Offers insights and discussions from industry experts. - **Must Read Research Papers**: Highlights important papers in the field. - **Company Tech Blogs**: Provides insights from tech companies on AI and machine learning. - **Practice Machine Learning**: Offers hands-on projects and exercises. - **ML in Production**: Guides practitioners

### Overview: Bolt is a high-performance, flexible deep learning library designed for efficient deployment across various platforms. It automates the deployment pipeline, achieving significant acceleration and supports a wide range of models and precision levels. ### Key Features: - **High Performance:** 15%+ faster than existing open-source acceleration libraries. - **Rich Model Conversion:** Supports Caffe, ONNX, TFLite, and Tensorflow. - **Various Inference Precision:** Supports FP32, FP16, INT8, and 1-BIT precision. - **Multiple Platforms:** ARM CPU (v7, v8, v8.2+, v9), X86 CPU (AVX2, AVX512), GPU (Mali, Qualcomm, Intel, AMD). - **Bolt is the first to support NLP and common CV applications.** - **Minimize ROM/RAM usage** - **Rich Graph Optimization** - **Efficient Thread Affinity Setting** - **Auto Algorithm Tuning** - **Time-Series Data Acceleration** ### Tech Stack: C++ ### Why It Matters: - **High Performance:** Significantly faster than existing open-source acceleration libraries. - **Rich Model Conversion:** Supports a wide range of models and precision levels. - **Multiple Platforms:** Supports ARM, X86, and GPU. - **Bolt is the first to support NLP and CV applications.** - **Minimize ROM/RAM usage** - **Rich Graph Optimization** - **Efficient Thread Affinity Setting** - **Auto Algorithm Tuning**

### Overview: This self-paced bootcamp on Generative AI provides tutorials on ML fundamentals, Ollama, LLMs, RAGs, LangChain, LangGraph, Fine-tuning, DSPy & AI Agents (CrewAI), and more. The curriculum includes real-world applications and covers Python essentials, mathematics for AI, and production systems. ### Key Features: - **Python Essentials for AI Engineering** - **Mathematics is the Language of AI** - **Start Simple - The Power of Linear Models** - **Essential PyTorch for Real-World Applications** - **Understanding Your Data - Data Exploration** - **MLOps and Production Systems** ### Tech Stack: Jupyter Notebook ### Why it Matters: This bootcamp equips you with the skills and knowledge to become a great AI engineer, focusing on real-world applications and practical, hands-on training.

Overview: DEvol (DeepEvolution) is a basic proof of concept for genetic architecture search in Keras, designed for classification problems. The current setup allows models to evolve in terms of convolutional and dense layers, with options for various parameters like feature maps, activation functions, dropout, and batch normalization. Key features: - Genetic architecture search for neural networks - Evolution of convolutional and dense layers with configurable parameters - Simple visualization of genetic crossover and mutation Tech stack: Python Why it matters: DEvol offers a novel approach to genetic architecture search, potentially accelerating the development of complex neural architectures. The simplicity of the setup and the ability to evolve models in a controlled environment make it a valuable tool for researchers and developers looking to explore new neural network architectures.

Overview: This repository serves as a comprehensive summary of diffusion model-based image processing, including restoration, enhancement, coding, and quality assessment. Key features: - Survey for diffusion model-based Image Restoration - Summary for diffusion model-based Image/Video Compression - Summary for diffusion model-based Quality Assessment Tech stack: Python, TensorFlow, PyTorch, and other related libraries. Why it matters: This repository aims to provide a detailed overview of recent advancements in diffusion model-based image processing, including restoration, enhancement, compression, and quality assessment, to help developers and researchers stay updated with the latest techniques and applications in the field. GitHub stars: 954 Repository: lixinustc/Awesome-diffusion-model-for-image-processing

### Overview: VideoComposer is an official repository for a controllable video diffusion model that allows users to control spatial and temporal patterns within a synthesized video. This model is designed to generate videos with various forms of input, including text descriptions, sketch sequences, reference videos, and even handcrafted motions and hand-drawn images. ### Key Features: - **Controllable Synthesis**: Users can control the spatial and temporal patterns of a synthesized video. - **Versatile Inputs**: Supports text descriptions, sketch sequences, reference videos, and handcrafted motions. - **Model Compatibility**: Compatible with various pretrained models and models from external projects like MiDaS, Open Clip, and Pidinet. ### Tech Stack: - Python - FFMpeg - Torch - torchvision - Open-Clip-Torch - Transformers - Flash-Attn - Xformers - Motion-Vector-Extractor ### Why It Matters: VideoComposer opens up new possibilities in video synthesis, enabling users to create videos that are not only visually appealing but also highly controlled and tailored to specific needs. This model can be particularly useful in applications such as video editing, content creation, and interactive storytelling. ### Repository: The repository for VideoComposer is hosted on GitHub at [ali-vilab/videocomposer](https://github.com/ali-vilab/videocomposer). It has been starred by 953 developers, indicating its popularity and utility in the community. ### Primary Language: The primary language used in the repository is Python. ### GitHub Stars: The repository has been starred by

Overview: This project involves the scraping of content from the Baidu Encyclopedia, extracting triples, and constructing a Chinese knowledge graph. Key features: - Scraping of Baidu Encyclopedia pages - Extraction of triples and webpage content - Construction of a Chinese knowledge graph Tech stack: Python Why it matters: This project is significant for its contribution to the development of Chinese knowledge graphs, providing a tool for understanding and analyzing Chinese knowledge through web scraping and graph database technology. Repository: lixiang0/WEB_KG Primary language: Python GitHub stars: 953 GitHub description: 爬取百度百科中文页面，抽取三元组信息，构建中文知识图谱 README excerpt:

Overview: Three Man Team is a structured 3-agent AI dev team designed for developers, providing clear roles and handoffs to prevent costly mistakes. Key features: - Arch generates a manifest file at first-time setup for easy install and management. - Improved version check reads a version registry instead of hitting GitHub API, making updates mandatory. - Arch reads actual files before walking through changes, reducing guessing. - Removed the VERSION file, keeping version in manifest.md. - Three Man Team is pre-built for developers, marketers, content creators, and more, with install times and productivity gains. - The team consists of an Architect, Builder, and Reviewer, each with distinct roles and responsibilities. - Quick start guide includes how to clone the repository, run setup, and integrate with Claude Code. Tech stack: Shell Why it matters: Three Man Team ensures a structured workflow, preventing costly mistakes and optimizing productivity by providing clear roles and handoffs. Repository: russelleNVy/three-man-team GitHub stars: 953 GitHub description: A structured 3-agent AI dev team — Architect, Builder, Reviewer. Built from production use. Token-optimized. Works with Claude Code, VS Code, Cursor, and any AI that supports context files.

Overview: This repository contains simple implementations of various natural language processing models, including search engines, word and sentence understanding, sequence-to-sequence models, attention mechanisms, and pretrained models like ELMo, GPT, and BERT. The tutorials are written in Chinese and can be found on my website. Key features: - TF-IDF implementation using numpy and sklearn - Skip-gram and Continuous Bag of Words (CBOW) models for word understanding - Sequence-to-sequence models for sequence generation - Attention mechanisms for improved model performance - Pretrained models for efficient language understanding Tech stack: Python Why it matters: This repository provides a practical introduction to NLP models and techniques, making complex concepts accessible through simple implementations. It serves as a valuable resource for developers and learners interested in deep learning and natural language processing.

### Overview: Selfhost AI is a one-command installer for a self-hosted AI stack on your own server, including n8n, Ollama, Open WebUI, OpenClaw, Dify, Flowise, Supabase, ComfyUI, and 30+ tools. It uses Docker Compose, automatic HTTPS via Caddy, and built-in monitoring. ### Key Features: - **One-Command Install**: An interactive wizard generates secrets and no manual configuration. - **Private AI Homelab**: Runs LLMs locally with Ollama, including multi-GPU support. - **ChatGPT Alternative**: Open WebUI for local and API models, with optional code execution sandbox. - **Workflow Automation**: n8n with 400+ integrations, scalable workers, and community workflows. - **AI Agents and RAG**: OpenClaw, Flowise, Keelflow, Dify, Letta, LightRAG, RAGFlow, Qdrant, Weaviate. - **Automatic HTTPS**: Caddy reverse proxy with Let's Encrypt; optional Cloudflare Tunnel. - **Built-in Monitoring**: Grafana and Prometheus with n8n workflow dashboard. - **Production Ready**: Health checks, service dependencies, and updates preserving settings. - **Free and Open Source**: Apache 2.0, no vendor lock-in. ### What's Included: - **Always Installed**: Caddy, PostgreSQL, and Valkey (Redis-compatible). - **Optional Services**: Open WebUI, OpenClaw, Open Terminal, Flowise

### Overview: OpenVINO Model Server (OVMS) is a scalable inference server optimized with OpenVINO™, designed for high-performance model serving across Generative AI, Agentic workloads, and classic deep learning models. It leverages Intel hardware for acceleration, making it suitable for deployment on various platforms including Linux, Windows, and Docker. ### Key Features: - **OpenAI-compatible API**: Supports text generation, embeddings, image generation, and audio. - **KServe APIs**: Provides classic model inference capabilities. - **Runs anywhere**: Deployed via Docker, bare metal, Kubernetes/OpenShift, and Windows. - **Intel-optimized**: Accelerated with CPU, GPU, and NPU via OpenVINO. ### Tech Stack: - **OpenVINO™**: Optimized inference server for models. - **C++**: Primary language for development and deployment. ### Why It Matters: OVMS is essential for deploying and serving machine learning models efficiently, ensuring optimal performance and reliability across different hardware architectures. Its ability to handle both Generative AI and classic deep learning models makes it a versatile tool for developers and researchers in the AI and machine learning field. ### Repository: The repository for OpenVINO Model Server is located at: [openvinotoolkit/model_server](https://github.com/openvinotoolkit/model_server). ### GitHub Stars: The repository has received 950 GitHub stars, indicating its popularity and utility in the community. ### Quick Start: #### Serve an LLM with OpenAI-compatible API: **On Linux (Docker):** ```bash

### Overview: Rhasspy is an offline voice assistant toolkit designed for home automation, supporting multiple languages including English, German, Spanish, French, Italian, Dutch, Russian, Greek, Hindi, Mandarin, Vietnamese, Portuguese, Swedish, Catalan, and more. It enables users to control their smart home devices with voice commands. ### Key Features: - **Offline Voice Commands**: Transcribes voice commands into JSON events for triggering Home Assistant automations or Node-RED flows. - **Profile Customization**: Supports custom voice commands defined in profiles using a specialized template syntax. - **Language Support**: Currently supports 24 languages, including English, German, Spanish, French, Italian, Dutch, Russian, Greek, Hindi, Mandarin, Vietnamese, Portuguese, Swedish, Catalan, and more. ### Tech Stack: - **Python**: The primary programming language used for Rhasspy. - **Docker**: Used for running the Rhasspy server. - **Home Assistant**: Integrates well with Home Assistant for seamless automation. - **Node-RED**: Supports Node-RED flows for more complex automation tasks. ### Why It Matters: Rhasspy simplifies the process of controlling smart home devices with voice commands, making it easier for users to manage their homes without the need for internet connectivity. Its versatility across multiple languages and integration with popular smart home platforms makes it a valuable tool for anyone looking to automate their living spaces.

### Overview: This repository contains a comprehensive list of papers and resources related to the application of Generative Artificial Intelligence (Generative AI) and Deep Learning for molecular design and molecular conformation generation. The focus is on using these advanced AI techniques to optimize molecular structures and create novel drugs and materials. ### Key Features: - **Unified Deep Molecular Generative Model** - **Generative AI for Scientific Discovery** - **Reviews** - **Datasets and Benchmarks** - **Drug-likeness and Evaluation metrics** - **Text-driven molecular generation models** - **Multi-Target based deep molecular generative models** - **Ligand-based deep molecular generative models** - **Pharmacophore-based deep molecular generative models** - **Structure-based deep molecular generative models** - **Fragment-based deep molecular generative models** - **Scaffold-based DMGs** - **Motifs-based DMGs** - **Linkers-based DMGs** - **Chemical Reaction-based deep molecular generative models** - **Omics-based deep molecular generative models** - **Multi-Objective deep molecular generative models** - **Quantum deep molecular generative models** - **Recommendations and References** ### Tech Stack: - Generative AI and Deep Learning ### Why It Matters: This repository serves as a valuable resource for researchers, developers, and anyone interested in the intersection of Generative AI and Deep Learning with molecular design and conformation generation. It provides a comprehensive overview of various approaches and techniques, making it an essential tool for anyone looking to explore and implement these innovative methods in

### Overview: Whisper.net is an open-source project that simplifies speech-to-text using Whisper Models, providing a straightforward and efficient solution for developers. ### Key Features: - **Speech-to-Text Integration**: Simplifies the process of converting spoken words into text. - **Native and Dotnet Bindings**: Compatible with both native and .NET environments, making it versatile for various development platforms. - **CUDA Support**: Available for GPU acceleration, enhancing performance on supported hardware. ### Tech Stack: C# ### Why it Matters: Whisper.net offers a user-friendly and efficient way to integrate speech recognition into applications, making it a valuable tool for developers looking to enhance their applications with natural language processing capabilities.

Overview: This initiative aims to compile and share papers and related summaries in a structured format, focusing on AI and technology advancements. Key features: - Papers on AI and technology advancements - Summaries and notes on each paper - Regular updates to the repository Tech stack: HTML Why it matters: This initiative helps developers stay updated with the latest in AI and technology, providing valuable insights and practical applications.

### Overview: 精选了10K+项目，涵盖机器学习、深度学习、NLP、GNN、推荐系统、生物医药、机器视觉、前后端开发等内容。让更多优秀的项目被人发现，让更多的人感受开源的魅力。 ### Key features: - 机器学习与深度学习 - NLP自然语言处理 * 大语言对话模型及数据 - 网络与前后端开发 - 机器视觉 - 语音识别与合成 - 推荐系统 - 因果推断 - 金融股票与时间序列 - 强化学习 - 生物医药 - 图数据库 图算法 - 图神经网络GNN - 大数据 - 虚拟化 - 安全与渗透 - 硬件 - 其他项目 ### Tech stack: Python, JavaScript, C++, Java, etc. ### Why it matters: 让更多优秀的项目被人发现，让更多的人感受开源的魅力。

Overview: CogView2 is an advanced text-to-image generator that utilizes hierarchical transformers to produce vivid and high-quality images from Chinese and English text. This implementation is based on the SwissArmyTransformer library and aims to provide faster and more effective text-to-image generation. Key features: - Hierarchical Transformer (6B-9B-9B parameters) for efficient text-to-image generation - Accelerates the generation process with LoPAR (Low-Precision Attention Residual) - Enables bidirectional completion with CogLM (Conditional Language Model) - Automatically downloads and integrates models into Huggingface Spaces and Replicate for easy access Tech stack: Python Why it matters: CogView2 offers significant improvements in text-to-image generation by leveraging hierarchical transformers, low-precision attention, and conditional language modeling, resulting in faster and more effective image generation compared to traditional methods. Repository: zai-org/CogView2 Primary language: Python GitHub stars: 952 GitHub description: official code repo for paper "CogView2: Faster and Better Text-to-Image Generation via Hierarchical Transformers"

Overview: This repository shares end-to-end examples of various Weaviate features and integrations, covering datasets, integrations, Weaviate features, and more. Key features: - Datasets - Integrations - Weaviate Features - Query Agent Tech stack: Jupyter Notebook Why it matters: This repository provides comprehensive guides on how to utilize Weaviate's features and integrate with other technologies, making it a valuable resource for developers looking to enhance their Weaviate projects.

### Overview: ReachAI 是一款面向企业级智能体开发与运行的平台，旨在快速、安全地将人工智能技术融入现有的业务系统中，从而提高工作效率并改善用户体验。 ### Key features: - **快速集成**：ReachAI 能够识别业务系统的技术栈，并生成标准接入任务，确保与业务系统的无缝连接。 - **安全运行**：平台通过身份传递和安全校验组件确保数据和业务操作的安全性。 - **智能推荐**：AI 主动扫描系统页面、接口和业务能力，推荐改造项，并生成标准化工程任务。 - **智能执行**：AI Coding 实施接入、改造和测试，确保代码的准确性和可追溯性。 - **全面验收**：真实业务验收确保了系统上线后的稳定性和可靠性。 - **持续改造**：通过项目管理、项目详情和持续改造闭环，实现业务系统的持续优化和升级。 ### Tech stack: ReachAI 使用 Java 作为主要开发语言，涵盖了前端、后端和安全等多个技术栈。 ### Why it matters: ReachAI 通过快速、安全的方式将人工智能技术融入企业系统，显著提升了企业的智能化水平，同时保证了数据的安全性和系统的稳定性，为企业带来显著的业务效益和效率提升。

### Overview: This repository is a curated collection of free, high-quality AI tools, APIs, datasets, and learning resources designed to help developers, researchers, and creators explore and build with AI faster. The resources are organized by topic and difficulty level, ensuring a structured and practical learning experience. ### Key Features: - **Structured Learning Paths:** Resources are organized by topic and difficulty level, making it easy to follow a learning path. - **Quality Verification:** Every resource is verified for free, high-quality, and from reputable sources. - **Community Driven:** Resources are maintained by a community, ensuring up-to-date information. - **Practical Paths:** Resources are categorized into beginner, intermediate, and advanced levels, making it easy to find the right starting point. ### Tech Stack: - Python - Machine Learning - Deep Learning - Natural Language Processing - Computer Vision - Generative AI - Data Science - AI Hardware - AI Agents - Prompt Engineering - AI Career Prep - AI in Healthcare - AI in Finance ### Why It Matters: This repository provides a comprehensive and structured approach to learning AI, making it easier for developers, researchers, and creators to build projects and advance their careers in the field. The resources are curated by a community, ensuring quality and up-to-date information.