Projects

### Overview: Rust-native state-of-the-art Natural Language Processing models and pipelines. This repository offers the base architecture, task-specific heads, and ready-to-use pipelines for various NLP tasks including translation, summarization, dialogue, classification, sentiment analysis, named entity recognition, part-of-speech tagging, question-answering, language generation, masked language modeling, and sentence embeddings. ### Key features: - **Translation**: Support for various NLP models including DistilBERT, MobileBERT, and more. - **Summarization**: Ready-to-use pipelines for summarization tasks. - **Multi-turn dialogue**: Models for handling multi-turn conversations. - **Zero-shot classification**: Models for zero-shot classification tasks. - **Sentiment Analysis**: Support for sentiment analysis. - **Named Entity Recognition**: Models for recognizing named entities. - **Part of Speech tagging**: Models for tagging parts of speech. - **Question-Answering**: Models for answering questions. - **Language Generation**: Models for generating language. - **Masked Language Model**: Models for masked language modeling. - **Sentence Embeddings**: Models for sentence embeddings. ### Tech stack: Rust ### Why it matters: This repository provides a Rust-native solution for NLP tasks, leveraging the power of Rust for performance and efficiency. It simplifies the integration of state-of-the-art NLP models and pipelines, making it easier for developers to implement and deploy NLP solutions in their projects.

Overview: Shogun is a machine learning toolbox that has been in use since 1999, offering unified and efficient solutions for various programming languages. Key features: - Python - Octave - Java/Scala - Ruby - C# - R - Perl - JavaScript (under development) - D (under development) - Matlab (under development) Tech stack: C++ Why it matters: Shogun is a powerful and versatile tool for machine learning, providing comprehensive solutions for developers across multiple programming languages, making it an essential resource for anyone working in the field of data science and artificial intelligence.

Overview: LlamaIndex.TS is a lightweight, TypeScript-based data framework designed for integrating large language models (LLMs) into applications with your own data. Key features: - Supports multiple JavaScript environments including Node.js, Deno, Bun, Nitro, and Vercel Edge Runtime. - Compatible with various LLM providers such as OpenAI, Anthropic, and others. - Provides a set of libraries for easy integration with large language models. - Includes a TypeScript documentation and examples for easy setup and usage. Tech stack: TypeScript Why it matters: LlamaIndex.TS simplifies the integration of large language models into applications, making it easier for developers to leverage their own data with these powerful AI tools. Repository: run-llama/LlamaIndexTS GitHub stars: 3082 GitHub description: Data framework for your LLM applications. Focus on server side solution.

### Overview: MemMachine is an open-source universal memory layer designed for AI agents and LLM-powered applications. It provides scalable, extensible, and interoperable memory storage and retrieval, enabling stateless agents to become personalized and context-aware assistants. ### Key Features: - **Episodic Memory**: Graph-based conversational context that persists across sessions. - **Profile Memory**: Long-term user facts and preferences stored in SQL. - **Working Memory**: Short-term context for the current session. - **Agent Memory Persistence**: Memory that survives restarts, sessions, and even model changes. ### Tech Stack: Python ### Why It Matters: MemMachine transforms stateless chatbots into personalized, context-aware assistants by providing persistent memory, enabling agents to learn, store, and recall information from past sessions. This technology is crucial for developing next-generation autonomous systems that require seamless and efficient state management.

### Overview: HeavyDB, formerly MapD/OmniSciDB, is an in-memory, column store, SQL relational database designed for GPU-accelerated performance. This repository now includes all functionality of the entire HeavyAI platform, including advanced visualization, GPU-accelerated rendering, simulation capabilities, and external data integration. ### Key Features: - **Immerse User Interface**: Advanced visualization and dashboard creation capabilities. - **HeavyAI Platform**: Includes a GPU-accelerated back-end Renderer that generates point/symbol/line/polygon/heatmap/raster images. - **HeavyRF**: Simulation capabilities for RF propagation. - **HeavyConnect**: Exposes external data through foreign tables for ETL workflows. - **Build Tools**: Simplified install process with Docker images. ### Tech Stack: - C++ ### Why It Matters: This project is now fully open source, providing all the functionality of the entire HeavyAI platform, including advanced visualization, GPU-accelerated rendering, and simulation capabilities. The inclusion of all features makes it a powerful tool for developers looking to leverage GPU performance for complex geospatial data analysis and visualization. ### Repository: - **GitHub Stars**: 3068 - **GitHub Description**: HeavyDB (formerly MapD/OmniSciDB) - **Repository**: heavyai/heavydb ### Quick Start Guide: For new users, the process of getting started is simple. Developers with GPU resources can quickly and easily get started either on-prem or in-cloud providers. Follow the instructions in [Build From Source](https://docs

### Overview: LimboAI is an open-source C++ plugin for Godot Engine 4 that provides Behavior Trees and State Machines, making it easier to create complex AI behaviors. ### Key features: - Behavior Trees and State Machines for Godot 4 - Behavior Tree Editor - Built-in Documentation - Visual Debugger - Extensive Demo Project with Tutorial - Custom Tasks and States Support ### Tech stack: C++ ### Why it matters: LimboAI simplifies the creation of complex AI behaviors in Godot, making it easier to implement rich and modular game mechanics. The plugin supports GDScript for creating custom tasks and states, enhancing flexibility and modularity.

### Overview: This repository is a curated list of deep learning image classification papers and codes, inspired by [awesome-object-detection](https://github.com/amusi/awesome-object-detection) and [deep_learning_object_detection](https://github.com/hoya012/deep_learning_object_detection). It aims to provide a comprehensive guide for beginners in deep learning, starting with VGG and moving on to more advanced models like GoogleNet, ResNet, and Inception series. The repository also includes a PyTorch implementation of some image classification networks. ### Key Features: - VGG - GoogleNet - PReLU-nets - ResNet - PreActResNet - Inceptionv3 - Inceptionv4 - Inception-ResNet-v2 - RiR - Stochastic Depth ResNet - WRN - SqueezeNet - GeNet - PyramidNet ### Tech Stack: PyTorch ### Why it Matters: This repository serves as a valuable resource for anyone starting their journey into deep learning, particularly focusing on image classification. It provides a structured way to learn and understand various models and their performance on ImageNet, helping beginners to make informed decisions about which models to study first.

### Overview: StellarGraph is a Python library designed for machine learning on graphs and networks, offering state-of-the-art algorithms for various tasks such as node and edge classification, attribute inference, graph classification, link prediction, and node interpretation. ### Key Features: - **Node and Edge Representation Learning**: Enables visualisation and downstream machine learning tasks. - **Node Classification and Edge Classification**: Identifies patterns and attributes within nodes and edges. - **Graph Classification**: Classifies entire graphs. - **Link Prediction**: Predicts the existence of links between nodes. - **Node Interpretation**: Provides insights into node classification through Saliency Maps. ### Tech Stack: Python ### Why it Matters: StellarGraph simplifies the process of applying machine learning techniques to graph-structured data, making it easier to discover patterns and answer questions about complex networked information. ### Repository: stellargraph/stellargraph ### GitHub Stars: 3064 ### Installation: - **Install StellarGraph using PyPI**: ```bash pip install stellargraph ``` - **Install StellarGraph in Anaconda Python**: ```bash conda install -c conda-forge stellargraph ``` - **Install StellarGraph from GitHub source**: ```bash git clone https://github.com/stellargraph/stellargraph cd stellargraph pip install . ``` ### Getting Help: - **Contribute**: Contributions are welcome and can be found in the [Contributions Guide](https://

### Overview: Tencent ML-Images is an open-source project that provides a large multi-label image database containing 17,609,752 training and 88,739 validation image URLs, with annotations up to 11,166 categories. The database is pre-trained on the ImageNet dataset and achieves a top-1 accuracy of 80.73% on ImageNet via transfer learning. ### Key Features: - **Large Multi-Label Database**: Contains 17,609,752 training and 88,739 validation image URLs. - **Annotations**: Up to 11,166 categories. - **Pre-trained on ImageNet**: Achieves 80.73% top-1 accuracy on ImageNet via transfer learning. ### Tech Stack: - **Python**: The primary programming language used for development. - **Tensorflow**: Used for training and pretraining the model. ### Why it Matters: Tencent ML-Images is significant for researchers and developers in computer vision and image classification tasks. It provides a comprehensive and annotated dataset that can be used for training and evaluating models, making it a valuable resource in the field of machine learning and artificial intelligence. ### Repository: - **Repository**: [Tencent/tencent-ml-images](https://github.com/Tencent/tencent-ml-images) - **Stars**: 3063 - **Description**: The largest open-source multi-label image database, including 17,609,752

Overview: We are developing an open-source dataset of chest X-ray and CT images for patients with suspected or confirmed COVID-19 or other respiratory conditions. Key features: - Publicly available chest X-ray and CT images of patients with suspected or confirmed COVID-19 or other respiratory conditions. - Annotations for lung bounding boxes and pneumonia severity scores. - Contributed datasets from General Blockchain, Brixia, and v7labs. Tech stack: Jupyter Notebook Why it matters: This dataset aims to provide a comprehensive resource for medical research and training in the field of respiratory imaging, particularly in the context of COVID-19. Repository: ieee8023/covid-chestxray-dataset GitHub stars: 3061

### Overview: Mini Agent is a minimal yet professional demo project that showcases best practices for building agents with the MiniMax M2.5 model. Leveraging an Anthropic-compatible API, it fully supports interleaved thinking to unlock M2's powerful reasoning capabilities for long, complex tasks. ### Key Features: - **Full Agent Execution Loop**: A complete and reliable foundation with a basic toolset for file system and shell operations. - **Persistent Memory**: An active **Session Note Tool** ensures the agent retains key information across multiple sessions. - **Intelligent Context Management**: Automatically summarizes conversation history to handle contexts up to a configurable token limit, enabling infinitely long tasks. - **Claude Skills Integration**: Comes with 15 professional skills for documents, design, testing, and development. - **MCP Tool Integration**: Natively supports MCP for tools like knowledge graph access and web search. - **Comprehensive Logging**: Detailed logs for every request, response, and tool execution for easy debugging. - **Clean & Simple Design**: A beautiful CLI and a codebase that is easy to understand, making it the perfect starting point for building advanced agents. ### Tech Stack: Python ### Why it Matters: Mini Agent provides a robust and intelligent agent development experience with features designed for a professional and efficient workflow. It is perfect for beginners and advanced users alike, ensuring a smooth and effective agent development process.

Overview: `prompttools` is an open-source project designed for developers to experiment with LLMs, vector databases, and prompts. It provides a set of tools that can be used to test and evaluate these technologies using familiar interfaces like code, notebooks, and a local playground. Key features: - A comprehensive set of experiments for LLMs, including OpenAI, Anthropic, and LLaMA models. - Support for vector databases such as Chroma, Weaviate, and LanceDB. - Easy setup and installation via `pip`. - A simple example script for testing prompts and parameters. - A playground interface for interactive experimentation. Tech stack: Python Why it matters: `prompttools` offers a convenient and accessible way for developers to evaluate and experiment with LLMs and vector databases, making it easier to understand and improve their models and database performance.

### Overview: EAST is a high-performance TensorFlow implementation of the EAST text detector, designed for accurate scene text detection. It includes an online demo for quick results and a pre-trained model achieving 80.83 F1-score on ICDAR 2015. ### Key Features: - **Online Demo**: Accessible at http://east.zxytim.com/, featuring a single CPU core demo server with degraded response time due to concurrency. - **RBOX Implementation**: Focuses on the RBOX part of the EAST model. - **Locality-Aware NMS**: Provided by the paper's author. - **Pre-trained Model**: Achieves 80.83 F1-score on ICDAR 2015 using only training images from ICDAR 2015 and 2013. - **Model Differences**: Uses ResNet-50 instead of PVANET, dice loss instead of balanced cross entropy, and linear learning rate decay instead of staged learning rate decay. - **Speed**: On 720p images, the model now runs at 16 fps with a GTX 1080 Ti graphic card, compared to 2 fps on a K40 with the original model. ### Tech Stack: The project is built using TensorFlow, a popular open-source machine learning framework. ### Why It Matters: This implementation is valuable for developers looking to improve their text detection models, especially in real-time applications where speed and accuracy are crucial. ### Repository: The repository is hosted on GitHub at https://github.com

### Overview: ALLWEONE® AI Presentation Generator is an open-source tool designed to create professional slides with customizable themes and AI-generated content. It offers a quick and easy way to generate presentations on any topic, with the ability to review and modify AI-generated outlines before finalizing. ### Key Features: - **AI-Powered Content Generation**: Create complete presentations on any topic with AI. - **Outline-First Workflow**: Generate an outline first, review it, then turn it into slides. - **Customizable Slides**: Choose the text model, number of slides, language, and whether web search is enabled. - **Blank Presentations**: Start from scratch when you do not want AI-generated content. - **Editable Outlines**: Review and modify AI-generated outlines before finalizing. - **Real-Time Generation**: Watch your presentation build live as content is created. - **Auto-Save**: Everything saves automatically as you work. ### Tech Stack: - **Framework**: Next.js, React, TypeScript - **Styling**: Tailwind CSS - **Database**: PostgreSQL with Prisma ORM - **AI Integration**: OpenAI API, Together AI, Ollama, LM Studio - **Authentication**: NextAuth.js - **UI Components**: Radix UI - **Text Editor**: Plate Editor - **File Uploads**: ### Why It Matters: This tool is part of the broader ALLWEONE AI platform, aiming to help developers reach more developers and grow the ALLWEONE community by providing an alternative to Gamma.app. The open-source nature and ease of use make

Overview: This repository offers a collection of Jupyter Notebooks designed to help developers get hands-on experience with Pinecone vector databases and various AI patterns, tools, and algorithms. Key features: - Production-ready examples - Learning-focused examples for AI techniques - Patterns for building applications Tech stack: Jupyter Notebook Why it matters: This repository provides practical, hands-on learning opportunities for developers to explore Pinecone's vector databases and AI-related concepts, making it an essential resource for developers looking to integrate Pinecone into their projects. Repository: pinecone-io/examples Primary language: Jupyter Notebook GitHub stars: 3050 GitHub description: Jupyter Notebooks to help you get hands-on with Pinecone vector databases README excerpt: 

Overview: QuantResearch is a comprehensive repository for quantitative research, including backtesting, machine learning, and live trading demos. It provides a range of notebooks and blogs for various quantitative techniques and strategies. Key features: - Backtesting - Machine Learning and Deep Reinforcement Learning - Live Trading Demo Video Tech stack: Jupyter Notebook Why it matters: QuantResearch offers a valuable resource for quantitative analysts, traders, and researchers looking to implement and test various quantitative strategies and backtesting techniques.

Overview: Spotlight is a deep learning project that utilizes PyTorch to develop recommender models. This project aims to enhance user experience by providing personalized recommendations based on their preferences. Key features: - Implementing advanced deep learning models for recommendation systems - Utilizing PyTorch for efficient model training and inference - Integrating real-time data processing capabilities Tech stack: Python with PyTorch Why it matters: Spotlight showcases the potential of deep learning in recommendation systems, offering a significant improvement in user satisfaction and engagement.

Overview: SWIRL is a federated AI search and RAG solution that allows developers to integrate AI-powered search capabilities into their applications without moving or duplicating data. It provides instant answers from company knowledge across 100+ apps while keeping data secure and accessible. Key features: - Federated search and RAG across 100+ apps - Live querying of sources with user permissions - Open source, running on your machine in about 2 minutes - Supports real-time RAG with OpenAI key - Galaxy UI for easy integration and querying Tech stack: Python Why it matters: SWIRL eliminates the need for data to be moved or duplicated, offering a secure and efficient way to integrate AI search capabilities into applications. It's open source and easy to deploy, making it a valuable tool for developers looking to enhance their applications with AI-powered search. Repository: swirlai/swirl-search GitHub stars: 3047

### Overview: Natively is a free, open-source AI meeting assistant that offers real-time transcription, AI meeting notes, lecture recording, local RAG, BYOK, and stealth mode. It provides a native macOS interface and is undetectable, making it an excellent alternative to Cluely, Otter, Granola, Final Round AI, Fireflies, and Interview Coder. ### Key features: - Real-time transcription - AI meeting notes - Lecture recording - Local RAG - BYOK (Bring Your Own Key) - Stealth mode - Source code available ### Tech stack: TypeScript ### Why it matters: Natively is a powerful tool for anyone looking for a free, undetectable AI assistant that can help with meeting notes, transcription, and more. Its native macOS interface and BYOK feature make it a standout alternative to other AI tools.

### Overview: Rebiber is a simple tool designed to update BibTeX entries with official information from sources like DBLP or the ACL Anthology. It ensures that cite keys remain unchanged while replacing unofficial BibTeX entries with accurate references. ### Key Features: 1. **Official Information**: Replaces unofficial BibTeX entries with official records from DBLP or the ACL Anthology. 2. **Single-Space Field Indentation**: Preserves single-space field indentation and one blank line between entries. 3. **Field Selection**: Allows customization of fields to be kept or removed. 4. **Title Protection**: Braces acronyms in titles to prevent conflicts. 5. **Sorting**: Sorts entries by cite key for better readability. ### Tech Stack: Python ### Why It Matters: Rebiber simplifies the process of updating BibTeX entries with official references, ensuring consistency and accuracy. It is particularly useful for researchers and scholars who need to cite sources accurately and efficiently.