Core Technologies
Computer vision, natural language processing, reinforcement learning and other AI core technologies
A stable diffusion web interface, developed based on the grado library, provides a friendly browser interface for users to visualize and operate the stable diffusion model conveniently.
An advanced natural language processing model library built for Jax, PyTorch and TensorFlow. It provides a rich set of pre-trained models and tools to help users achieve better results and performance in natural language processing tasks.
Netdata - Netdata can perform distributed, real-time performance and health monitoring for systems or applications
Microsoft Azure cloud advocates are pleased to offer a 12-week, 26-lesson machine learning course. In this course, you will learn what is sometimes called classical machine learning, using Scikit-learn as the library, avoiding deep learning, which will be covered in our upcoming "Beginner AI" course. Pair these courses with our "Beginner Data Science" course!
A free online open source textbook "Hands-On Deep Learning" is a Chinese translation of the Introduction to Deep Learning course textbook at the University of California, Berkeley for the spring semester of 2019. The book comprehensively introduces deep learning, not only explaining the principles of algorithms, but also running them to gain an interactive learning experience.
A selected list of computer science video courses covering areas such as introduction to computer science, data structures and algorithms, system programming, software engineering, artificial intelligence, machine learning, etc., including open courses from famous universities and professional lecturer courses.
scikit-learn is a Python module for machine learning built on top of SciPy.
A project dedicated to large-scale system design, which gathers the patterns and best practices of scalable, reliable and high-performance systems. It provides developers with rich resources and references to help them design and implement efficient large-scale systems.
The world's simplest facial recognition library
This is a tool that uses deep learning technology to identify and exchange faces in pictures and videos. It uses neural network models, by learning a large amount of face data, it can accurately identify the faces in images or videos and replace them with other faces. This tool has powerful face recognition and exchange capabilities, which can realize various interesting applications, such as replacing one person's face with another person's face, or realizing face expression and mouth shape conversion in videos.
A collection of articles covering fields such as graph neural networks, large language models (LLMs), and convex optimization. The project provides in-depth technical discussions, covering the principles of graph neural networks, applications of large language models, and the importance of convex optimization in relevant areas.
A project called Colossal-AI, which is open-sourced by Chinese people on GitHub, only needs a small amount of modification to enable existing deep learning projects to complete large model training on a single consumer-level graphics card, greatly reducing the cost of project development! In short, with this open-source project, everyone can train AI large models at home! Especially, it has significantly reduced the threshold for fine-tuning, inference and other downstream tasks and application deployment of AI large models.
Ultralytics provides a new deployment tool that supports the conversion of YOLOv8 models from PyTorch to various platforms, including ONNX, OpenVINO, CoreML and TFLite. For computer vision developers and researchers, this tool can help quickly deploy YOLOv8 models to different platforms, improving the performance of model applications.
A scheduled task management platform, which manages and schedules various offline scheduled tasks with a built-in web management interface. When the number of scheduled tasks reaches hundreds, it becomes impossible to effectively and conveniently manage these tasks using crontab. This project was born to solve this problem.
TTS is a library for advanced text-to-speech generation. It includes pre-trained models in over 1100 languages. Tools are provided for training new models and fine-tuning existing ones for any language. Utility programs are available for dataset analysis and management.
Streamlit is an open-source Python library that makes it easy to create and share beautiful custom web applications for machine learning and data science. Streamlit converts data scripts into sharable web applications in minutes. It's all written in pure Python. No front-end experience is required, so you can build and share data applications faster than ever before.
The open-source project named Gradio on GitHub can generate a simple and elegant UI interface for machine learning models in just a few minutes, allowing you to demonstrate your projects in the browser. Through this interface, you can complete operations such as dragging and uploading images, pasting text, recording sounds, etc., and view the model output content.
Chat-based large language models can interact with third-party systems and dynamically retrieve information.
Microsoft Azure cloud advocates are pleased to offer a 12-week, 24-lesson artificial intelligence course. In this course, you will learn: different approaches to artificial intelligence, including the "good old" symbolic methods of knowledge representation and reasoning (GOFAI). Neural networks and deep learning, which are at the core of modern artificial intelligence. We will use code from two of the most popular frameworks - TensorFlow and PyTorch - to illustrate the concepts behind these important topics. Neural architectures for processing images and text. We will introduce the latest models, but may lack some of the most advanced. Less popular artificial intelligence methods, such as genetic algorithms and multi-agent systems.
A decentralized web AI photo app🌈💎✨. It provides powerful features including photo management, smart categorization, search and sharing. Photoprism uses AI technology to automatically recognize objects, scenes and people in photos, allowing you to easily organize and browse your photo collection.
A practical real-world face restoration algorithm. It utilizes the rich prior knowledge from pre-trained face GANs (such as StyleGAN2) for blind face restoration, improving the quality and realism of the restored faces.
An open-source PyTorch tutorial code, the content is mainly tutorial code provided for deep learning researchers to learn PyTorch. Most of the models are implemented with less than 30 lines of code. You can use it in conjunction with the official PyTorch reference for learning.
Netron is a versatile viewer for neural network, deep learning, and machine learning models, supporting a wide range of formats including ONNX, TensorFlow Lite, Core ML, Keras, Caffe, Darknet, PyTorch, and more. It offers experimental support for additional frameworks like TorchScript, OpenVINO, and PaddlePaddle. Available across multiple platforms—macOS, Linux, Windows, browsers, and Python—Netron provides an intuitive interface for visualizing and analyzing model architectures. Users can easily install it via package managers or directly download the application. Sample models are provided for quick exploration, making it a valuable tool for developers and researchers working with diverse machine learning frameworks.
It selects about 10 influential articles in the field of deep learning in the past decade, and provides video explanations for them. The articles will cover common frameworks, algorithms, and practical applications in the field of deep learning, such as ResNet, Transformer, BERT, GPT3, GAN, AlphaGo, etc.
An AI technology roadmap, initiated by the German software company AMAI GmbH, contains relevant knowledge points in the field of AI technology, each of which is accompanied by detailed documents
A neural network based on multimodal (image, text) contrastive training. It can predict the most relevant text snippets using natural language given an image, without the need for optimization for specific tasks. CLIP is designed similarly to GPT-2 and GPT-3, boasting excellent zero-shot capabilities and can be applied to a variety of multimodal tasks.
This project allows real-time face swapping in PC streaming or video calls. It provides an interesting and innovative way to interact with and entertain faces.
A selection of papers, technical articles and well-known blogs related to data science and machine learning, covering 24 technical directions such as data engineering, natural language processing, computer vision, reinforcement learning, etc. Most of the articles come from world-renowned universities and enterprises.
A community-supported supercharged version of paperless: scan, index, and archive all your physical documents.
A project for writing reusable computer vision tools. Through this project, users can more easily create and manage the tools and processes needed for computer vision applications. Whether it's dataset preparation or model training, Supervision provides tools to help developers.
Make it more efficient and convenient for Python developers to integrate image text recognition functions
Develop voice applications: Alexa, Google Home, Siri, Cortana, and Mycroft Mark II.
An interactive deep learning book that provides code, math, and discussions across multiple frameworks. This project has been adopted at over 500 universities in 70 countries around the world, including Stanford University, Massachusetts Institute of Technology, Harvard University, Cambridge University, etc. It provides rich resources and an interactive learning experience for learning deep learning.
Open source free machine learning book "PumpkinBook", which is a subtle supplement to the classic introductory textbook of machine learning by Prof. Zhichao Zhou, such as analyzing difficult formulas and supplementing specific derivation details
The main content of this book is the author's study during his graduate school period. The tutorial will be presented in the form of videos, mainly introducing the network structure and innovation points, as well as building and training networks based on PyTorch / TensorFlow.
An open platform for training, servicing, and evaluating large language models.
A very comprehensive algorithm resource, mainly divided into several categories such as supervised learning, unsupervised learning and neural networks, and provides the principle introduction and demo implementation of related algorithms
A vector database for next-generation AI applications. It provides efficient vector indexing and retrieval functions, supporting fast similarity search and relevance calculation, suitable for various AI application fields.
The non-profit technology organization fast.ai recently opened its new version of the deep learning course.
I have collected a lot of tutorials and example codes related to computer vision, deep learning, and artificial intelligence.
A 100-day machine learning tutorial in Chinese
Perplexica is an open-source, AI-powered search engine designed to deliver precise, up-to-date answers by leveraging advanced machine learning techniques like similarity searching and embeddings. It integrates with SearxNG to ensure privacy and real-time information retrieval. Key features include support for local LLMs (e.g., Llama3, Mixtral via Ollama), two search modes (Normal and Copilot), and six focus modes tailored for specific needs like academic research, YouTube, Reddit, and Wolfram Alpha queries. Perplexica offers an API for seamless integration into applications and supports Docker for easy installation. It prioritizes user privacy and provides cited sources for transparency, making it a versatile tool for enhanced web searching.
The goal of the "PyTorch Chinese Manual" is to help friends who want to learn and use PyTorch for deep learning development and research get started quickly.
An open-source organization that provides C language implementations of various fundamental algorithms and data structures. The project includes sample code for basic algorithms, covering multiple programming languages, offering valuable resources for learning and understanding algorithms.
It includes some practical machine learning and Python open source projects and tools. There are more than 900 projects in total, including data visualization, natural language processing, text and image data, web crawling, etc.
An AI quantitative trading platform, aiming to seek and explore the value of AI technology in the quantitative investment field. It contains a complete ML pipeline for data processing, model training, and backtesting, and covers the entire chain of quantitative investment: Alpha exploration, risk modeling, portfolio optimization, and transaction execution functions.
Convert the MXNet code implementation in the original "Dive into Deep Learning" book to a PyTorch implementation.
A practical list of machine learning projects, which includes a series of excellent open-source machine learning libraries that can help you monitor, deploy and protect your project.
Developers can replace the face and avatar in the video at will, as well as match the mouth
The Llama Cookbook is the official guide for building applications with Llama Models, offering comprehensive resources for inference, fine-tuning, and end-to-end use cases. It includes tutorials for popular models like Llama 4 Scout and Llama 3.2 Vision, as well as safety-focused tools like Llama Guard. The repository features practical recipes for diverse applications, such as email agents, text-to-SQL, and multimodal inference. Structured into sections for third-party integrations, use cases, and getting started, it also provides FAQs and fine-tuning guidance. The project supports community contributions and adheres to Meta’s licensing and acceptable use policies for Llama models.
A project to accelerate the training of NeRF models. NVIDIA's open source technology can train a fox NeRF model in just 5 seconds, which is based on static 2D images and uses neural networks to quickly train pictures that can be magnified and clearly observed from any angle.
The ML YouTube Courses project is dedicated to providing users with the latest machine learning and artificial intelligence courses, all of which can be found on YouTube. By aggregating various educational resources, this project offers learners and practitioners a convenient platform to easily browse, filter, and select course content that suits their learning needs. Whether you are a beginner or a professional, ML YouTube Courses is an ideal choice for discovering quality machine learning educational resources.
A cross-platform, high-performance machine learning inference and training accelerator launched by Microsoft. The project aims to provide machine learning developers with a flexible, high-performance runtime environment that supports multiple frameworks and hardware accelerators, making model inference and training more efficient. ONNX Runtime not only supports multiple operating systems but can also seamlessly integrate into various applications and services, providing convenience for the deployment of machine learning applications.
