Projects

### Overview: This repository is a comprehensive collection of free resources designed to help developers learn everything about AI/ML, LLMs, and Agents. It covers foundational mathematics, Python, AI/ML fundamentals, machine learning frameworks, deep learning frameworks, specializations, generative AI, and large language models. ### Key Features: - **Mathematics Roadmap for Machine Learning**: A detailed roadmap for understanding the mathematical foundations of machine learning. - **Python**: Essential resources for beginners to AI/ML. - **AI & ML Fundamentals**: Comprehensive guides and courses on AI and machine learning. - **Machine Learning Frameworks**: TensorFlow, PyTorch, and Keras. - **Deep Learning**: TensorFlow, PyTorch, and Keras. - **Deep Learning Specializations**: Computer Vision, NLP, and Reinforcement Learning. - **Generative AI**: Resources for understanding and building generative AI models. - **Large Language Models (LLMs)**: Explained through visual guides and YouTube videos. ### Tech Stack: - Python - TensorFlow - PyTorch - Keras - Scikit-learn - XGBoost - LightGBM - CatBoost - Scikit-learn - XGBoost - LightGBM - CatBoost ### Why it Matters: This repository serves as a valuable resource for developers aiming to build a strong foundation in AI/ML, LLMs, and Agents. It provides a structured approach to learning, making it accessible and comprehensive for beginners and intermediate learners alike.

### Overview: Cactus is a cutting-edge AI engine designed for mobile devices, wearables, smart home, and robots. It leverages zero-copy computation graphs, custom quantization techniques, and OpenAI-compatible APIs for seamless integration with various AI models. ### Key Features: - **Cactus Engine**: A powerful hybrid edge-cloud AI engine capable of handling text, speech, and vision tasks. - **Cactus Graph**: Enables zero-copy computation graphs, optimizing performance and resource efficiency. - **Cactus Kernels**: High-performance CPU/GPU kernels for devices like Apple, Samsung, and Pixel. - **Cactus Quants**: Custom rotation-based quantization technique for efficient inference. ### Tech Stack: - **C++**: The primary programming language used for development. ### Why It Matters: Cactus is essential for developers looking to integrate advanced AI capabilities into their mobile and wearable devices, enabling real-time applications that require high performance and low latency. Its ability to seamlessly integrate with OpenAI-compatible APIs and its optimized zero-copy computation graphs make it a valuable tool for edge computing and AI-powered devices.

### Overview: OpenCVSharp is a cross-platform .NET wrapper for OpenCV, providing a rich set of image processing and computer vision functionality. Its API is designed to closely mirror the native OpenCV C++ API, making it easy to integrate with existing C++ codebases while still adhering to C# conventions. ### Key Features: - OpenCVSharp provides a comprehensive set of image processing and computer vision functionalities. - It supports both .NET 8+ and .NET Framework 4.6.1+ targets. - The package layout mirrors OpenCV 5.x, ensuring compatibility and ease of migration. - It includes a live demo in Blazor WebAssembly, allowing quick experimentation without installation. ### Tech Stack: - .NET 8+ and .NET Framework 4.6.1+ targets ### Why It Matters: - OpenCVSharp simplifies the integration of OpenCV functionalities into .NET applications, making it easier to leverage existing C++ codebases. - The package layout ensures minimal friction when transferring OpenCV concepts and code to C#. ### Repository: - Repository: [shimat/opencvsharp](https://github.com/shimat/opencvsharp) - GitHub Stars: 6113 - GitHub Description: OpenCV wrapper for .NET ### Installation: - For Windows (x64): `dotnet add package OpenCvSharp5` - For Windows (ARM64): `dotnet add package OpenCvSharp5.runtime.win-arm64` - For Linux / Ubuntu: `dotnet add package OpenC

Overview: The Agents CLI is a powerful tool designed to transform any coding assistant into an expert in creating, evaluating, and deploying AI agents on Google Cloud. Key features: - `agents-cli setup`: Installs the CLI and necessary skills for coding agents. - `agents-cli create`: Creates new agent projects. - `agents-cli eval run`: Runs the agent over the evaluation dataset and grades the traces. - `agents-cli deploy`: Deploys the agent to Google Cloud. - `agents-cli publish gemini`: Registers the agent with the Gemini platform. Tech stack: Python, Node.js, and Google Cloud. Why it matters: It simplifies the process of building, deploying, and managing AI agents, making it easier for developers to leverage Google Cloud's powerful AI services without needing to learn every CLI and service themselves.

Overview: Higgsfield is an open-source, fault-tolerant, and highly scalable GPU orchestration framework designed for training models with billions to trillions of parameters, such as Large Language Models (LLMs). Key features: - Fault-tolerant GPU orchestration - Efficient sharding for trillion-parameter models - Framework for initiating, executing, and monitoring large neural network training - Resource management for continuous integration and seamless integration with GitHub and GitHub Actions Tech stack: Jupyter Notebook Why it matters: Higgsfield provides a robust and versatile toolset for training massive models, ensuring efficient resource management and seamless integration with popular tools and platforms. Repository: higgsfield-ai/higgsfield GitHub stars: 6072 GitHub description: Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters, such as Large Language Models (LLMs).

Overview: This project, sensitive-word, is a Java-based framework designed for detecting and filtering sensitive words. It leverages DFA algorithm for high performance and supports various features such as word classification, support for multiple character encodings, and dynamic data updates. The project aims to provide a robust and efficient solution for sensitive word detection in Java applications. Key features: - High performance DFA algorithm implementation - Support for fluent API for easy usage - Large 6W+ word library with continuous updates - Support for various formats conversion - Real-time data dynamic update - Support for custom sensitive words and whitelists - Word matching modes with two options Tech stack: Java Why it matters: This project is crucial for developers looking to add sensitive word detection capabilities to their Java applications, ensuring compliance with regulatory requirements and maintaining a clean, secure environment. Repository: houbb/sensitive-word Primary language: Java GitHub stars: 6077 GitHub description: The sensitive word tool for Java. Based on DFA algorithm implementation, it provides a high-performance sensitive word detection filter component with advanced features like word classification, support for multiple character encodings, and real-time data updates. It also includes a large 6W+ word library and supports various format conversions and dynamic data updates. The project aims to help developers ensure compliance with regulatory requirements and maintain a clean, secure environment. The sensitive-word-admin project, sensitive-word-admin, is a console application that provides a more flexible and convenient way to configure sensitive word detection. It is currently in the initial development phase and can be

### Overview: The **Tree of Thoughts (ToT)** project is a comprehensive implementation of the NeurIPS 2023 paper "Tree of Thoughts: Deliberate Problem Solving with Large Language Models". This project provides a framework for using large language models to solve problems through deliberate problem-solving techniques, including game solving, creative writing, and more. The project includes code, prompts, and model outputs, as well as a tweet thread detailing the experiment. ### Key Features: - **Game Solving**: Solves the game of 24 using a GPT-4 model. - **Creative Writing**: Generates thought sequences for creative writing tasks. - **Model Outputs**: Provides detailed outputs of the model's thought processes and state evaluations. ### Tech Stack: Python ### Why It Matters: This project showcases the capabilities of large language models in solving complex problems and provides a practical application of deliberate problem-solving techniques. The use of GPT-4 in the game of 24 task demonstrates the model's ability to generate coherent and logical solutions, while the creative writing task highlights the model's potential for generating thoughtful and engaging content. ### Repository: The project is hosted on GitHub at **princeton-nlp/tree-of-thought-llm**. It has been starred by 6,078 users, indicating high interest and engagement. ### Primary Language: The project is written in Python. ### GitHub Stars: The project has received 6,078 GitHub stars, indicating significant interest and community support. ### Repository Description: The

### Overview: Remove AI Watermarks is a powerful tool designed to eliminate visible and invisible AI watermarks from images, videos, and audio. This library is essential for developers and content creators who want to ensure the authenticity and provenance of their work. ### Key Features: - **CLI Tools**: `uv tool install "remove-ai-watermarks[visible]"` for visible watermark removal and metadata stripping. - **Agent Skill**: `npx skills add wiltodelta/remove-ai-watermarks` for Claude Code, Codex, Cursor, and other agents. - **Video Support**: Removes visible marks and metadata from images, videos, and audio. Supports complete visible-plus-metadata cleaning, directory batches, and video SynthID removal. - **GPU Support**: The library is optimized for performance, making it suitable for both CPU and GPU environments. ### Tech Stack: Python ### Why it Matters: This tool is crucial for maintaining the integrity and provenance of your digital content, ensuring that your work is not tampered with or attributed to unauthorized sources. It is essential for developers and content creators who want to protect their intellectual property and maintain the authenticity of their creations. ### Repository: Repository: [wiltodelta/remove-ai-watermarks](https://github.com/wiltodelta/remove-ai-watermarks) ### GitHub Stars: 6,039 ### Primary Language: Python ### Why it Matters: This tool is crucial for maintaining the integrity and provenance of your digital content, ensuring that your work is not tampered with or attributed to unauthorized sources. It is essential

Overview: This project is a command-line interface tool designed to correct the writing style of AI-generated Korean text, making it sound more natural and human-like. It detects and corrects common stylistic and grammatical issues such as translationese, mechanical parallelism, and other AI-specific language patterns. Key features: - Corrects AI-generated Korean text to sound more natural and human-like. - Detects and corrects common stylistic and grammatical issues. - Supports multiple AI models like Claude, Codex, and Gemini. - Provides detailed reports and logs for debugging and performance analysis. Tech stack: Python Why it matters: This tool is crucial for improving the readability and authenticity of AI-generated content, making it more suitable for human consumption and enhancing the overall quality of AI-generated text. Repository: epoko77-ai/im-not-ai GitHub stars: 6030 GitHub description: AI가 쓴 한글을 사람 글처럼 윤문하는 Claude 스킬 — Korean AI-text humanizer: detects and rewrites translationese, mechanical parallelism, and 71 other AI tells

### Overview: VLM-R1 is a stable and generalizable R1-style Large Vision-Language Model designed to solve Visual Understanding tasks. It has achieved remarkable performance, particularly in Referring Expression Comprehension (REC) and Object-Verb Dependency (OVD) tasks, surpassing state-of-the-art results on both in-domain and out-of-domain datasets. ### Key Features: - **Full Fine-tuning for GRPO**: Supports fine-tuning for GRPO tasks. - **Freeze Vision Modules**: Allows freezing of vision modules for better performance. - **LoRA Fine-tuning for GRPO**: Enables LoRA fine-tuning for improved performance. - **Multi-node Training**: Supports multi-node training for better model efficiency. - **Multi-image Input Training**: Supports multi-image input training for more comprehensive understanding. - **For Your Own Data**: Provides support for training on your own data. - **Various VLMs**: Supports multiple vision-language models, including QwenVL and InternVL. ### Tech Stack: - Python ### Why It Matters: VLM-R1 stands out by providing stable and generalizable performance across various tasks, particularly excelling in tasks like Referring Expression Comprehension (REC) and Object-Verb Dependency (OVD). Its ability to generalize reasoning across different domains makes it a valuable tool in the field of visual understanding and language modeling.

Overview: Chinese-CLIP is a Chinese version of the CLIP model, designed to facilitate quick implementation of cross-modal retrieval, representation generation, and other tasks in the Chinese domain. It leverages large-scale Chinese data for training (~2 billion image-text pairs) and is built on the open-source project open_clip, aiming to optimize performance for Chinese-specific data. Key features: - Cross-modal retrieval - Zero-shot image classification - Textual representation generation Tech stack: Python (Jupyter Notebook) Why it matters: Chinese-CLIP simplifies the process of implementing cross-modal tasks in the Chinese language, making it easier for developers to integrate CLIP functionalities into their projects.

Overview: This project provides a comprehensive countdown system for various AI, CV, NLP, and ML conferences, helping researchers stay updated on deadlines. Key features: - A detailed list of top-tier AI conferences, including deadlines and locations. - Easy contribution process for adding or updating conference information. - Support for multiple time zones and languages. Tech stack: JavaScript Why it matters: This tool helps researchers and professionals keep track of important deadlines, ensuring they can submit their work on time and contribute effectively to the field of AI and related disciplines. Repository: paperswithcode/ai-deadlines GitHub stars: 6009 Primary language: JavaScript Contributing: Contributions are very welcome and can be made by forking the repository and updating the `_data/conferences.yml` file with conference details.

### Overview: Shimmy is a pure-Rust WebGPU inference engine that runs on any GPU, compatible with OpenAI APIs and GGUF models. It is a single-binary, 100% Rust project with no Python or llama.cpp dependencies. It is free forever and sponsored for certification, compatibility, and releases. ### Key Features: - **Single-Binary**: Runs on any GPU, no Python or llama.cpp dependencies. - **OpenAI-Compatible**: Compatible with OpenAI APIs. - **GGUF Native Support**: Native support for GGUF models. - **Pure-Rust Engine**: Pure-Rust WebGPU (WGSL) engine. - **Model Certification**: 26 models certified across 12 families. - **Model Spec Auto-Generated**: Model specifications derived from GGUF metadata. - **YaRN RoPE Scaling**: Extended context via SHIMMY_MAX_CTX. - **F32 Accumulation Precision**: Deterministic output with F32 accumulation precision. ### Tech Stack: Pure-Rust WebGPU (WGSL) ### Why It Matters: - **No Python or C++ Toolchain**: Pure-Rust, top-to-bottom. - **F32 Precision**: Accurate and deterministic output. - **WebGPU Compatibility**: Supports NVIDIA, AMD, Intel, and Apple Silicon GPUs. - **Model Spec Auto-Generated**: No hardcoded per-model constants. - **YaRN RoPE Scaling**: Extended context via SHIMMY_MAX_CTX. ### Repository: Michael-A-Kuykendall/shimmy ### GitHub Stars: 5982

Overview: Flash Linear Attention is an efficient implementation of emerging model architectures, providing hardware-efficient building blocks, training-ready layers, and components for modern sequence models. Key features: - Linear Attention - Sparse Attention - State Space Models - Hybrid LLM Architectures Tech stack: Python Why it matters: Flash Linear Attention offers platform-agnostic implementations verified on NVIDIA, AMD, and Intel hardware, making it suitable for various computing environments. Repository: fla-org/flash-linear-attention Primary language: Python GitHub stars: 5940 GitHub description: 🚀 Efficient implementations for emerging model architectures README excerpt: <div align="center"> <img width="50%" alt="Flash Linear Attention" src="images/logo.png"> <br> [](https://huggingface.co/fla-hub) [](https://discord.gg/vDaJTmKNcS) </div> <p> 💥 Flash Linear Attention brings together hardware-efficient building blocks, training-ready layers, and components for modern sequence models, spanning linear attention, sparse attention, state space models, and hybrid LLM architectures. All implementations are platform-agnostic and verified on NVIDIA, AMD, and Intel hardware. Pull requests are welcome! </p>

### Overview: Lemonade is a local AI server that provides access to the latest models for chat, coding, speech, and image generation, all without the need for cloud APIs. It optimizes performance for NVIDIA and AMD GPUs and NPUs, making it suitable for both personal and professional use. ### Key Features: - **Lemonade Server**: A service that integrates with popular OpenAI, Anthropic, and Ollama APIs. - **Embeddable Lemonade**: A portable binary that can be packaged into applications for local AI capabilities. - **Multi-Modal Support**: Supports chat, image generation, speech generation, and more. - **Optimized for NPU and GPU**: Runs efficiently on NVIDIA and AMD GPUs and NPUs. - **Community Built**: Optimized by AMD engineers to maximize performance on Ryzen AI, Radeon, and Strix Halo PCs. ### Tech Stack: C++ ### Why it Matters: Lemonade offers a 100% free and private alternative to cloud-based AI, providing the same capabilities as cloud APIs but without the cost and security risks. It is ideal for developers and businesses looking to integrate AI into their applications without the need for cloud infrastructure. ### Repository: lemonade-sdk/lemonade ### GitHub Stars: 5921 ### GitHub Description: Join our discord: https://discord.gg/5xXzkMu8Zk

### Overview: GluonCV is a comprehensive toolkit for computer vision research and development, designed to simplify the prototyping and implementation of state-of-the-art deep learning models. ### Key features: - **Training Scripts**: Replicates SOTA results from research papers. - **Supports PyTorch and MXNet**: Compatible with both frameworks. - **Large Pre-trained Models**: Offers a variety of pre-trained models for various tasks. - **Carefully Designed APIs**: Simplifies model implementation with reduced complexity. - **Community Support**: Provides resources and tutorials for developers. ### Tech stack: Python ### Why it matters: GluonCV is essential for engineers, researchers, and students looking to quickly prototype and research computer vision models. Its comprehensive features and community support make it a valuable tool in the field of deep learning and computer vision. ### Repository: dmlc/gluon-cv ### GitHub stars: 5913 ### GitHub description: Gluon CV Toolkit

Overview: GPUStack is an open-source GPU cluster manager designed for high-performance AI model serving and on-demand SSH-accessible GPU instances. It supports multiple environments, including on-premises, Kubernetes, and cloud providers, and offers pluggable inference engines for various models. Key features: - Multi-Cluster GPU Management - Pluggable Inference Engines - Day 0 Model Support - Performance-Optimized Configurations - GPU Instances - Enterprise-Grade Operations Tech stack: Python Why it matters: GPUStack simplifies the deployment and management of AI models, providing efficient GPU resource utilization and optimized performance, making it ideal for high-performance computing and model serving environments.

### Overview: This repository showcases a collection of recent diffusion models designed for video generation, editing, and various applications. It includes models and tools for tasks such as video generation, editing, and understanding, providing a comprehensive overview of the latest advancements in the field. ### Key features: - **Video Generation**: Models for generating videos from text descriptions or images. - **Efficient Video Generation**: Techniques for optimizing video generation processes. - **Controllable Video Generation**: Tools for controlling and manipulating video content. - **Character Customization**: Models for creating personalized characters in videos. - **Motion Customization**: Enhancing video motion for better visual effects. - **Long Video / Film Generation**: Models for creating long and complex videos. - **Video Generation with 3D/Physical Prior**: Incorporating 3D models and physical constraints in video generation. - **Video Editing**: Tools for editing and manipulating video content. - **Human or Subject Motion**: Techniques for enhancing human or subject motion in videos. - **Video Enhancement and Restoration**: Methods for improving video quality and restoration. - **Audio Synthesis for Video**: Combining audio and video for more immersive experiences. - **Talking Head Generation**: Models for generating talking heads in videos. - **Reinforcement Learning for Video Generation**: Using reinforcement learning to generate videos. - **Policy Learning**: Techniques for learning policies for video generation. - **Virtual Try-On**: Models for creating virtual try-on experiences. - **3D**: Models for generating 3D videos. - **4D**: Models for generating 4D videos. - **

Overview: Daft is a high-performance data engine designed for AI and multimodal workloads, processing images, audio, video, and structured data at any scale. Key features: - Native multimodal processing - Built-in AI operations (LLM prompts, embeddings, and data classification) - Python-native, Rust-powered - Seamless scaling (local, distributed clusters) - Universal connectivity (S3, GCS, Iceberg, Delta Lake, Hugging Face, Unity Catalog) - Out-of-box reliability (intelligent memory management, sensible defaults) Tech stack: Rust Why it matters: Daft provides a powerful and efficient solution for AI and multimodal applications, making it suitable for various industries and use cases. Repository: Eventual-Inc/Daft Primary language: Rust GitHub stars: 5806 GitHub description: High-performance data engine for AI and multimodal workloads. Process images, audio, video, and structured data at any scale.

### Overview: The Ghidra MCP Server is a comprehensive toolset for AI-powered reverse engineering, featuring over 200 MCP tools, a GUI plugin, and a headless server. It leverages Ghidra's powerful reverse engineering capabilities and integrates with modern AI workflows, including batch operations, lazy tool loading, convention enforcement, and Ghidra Server integration. ### Key Features: - **200+ MCP Tools:** A wide range of tools for AI-driven reverse engineering. - **GUI Plugin:** A user-friendly interface for easy access to MCP tools. - **Headless Server:** A lightweight server for running tools without a graphical interface. - **Lazy Tool Loading:** Tools are loaded only when needed, improving performance. - **Convention Enforcement:** Ensures consistency in tool usage and results. - **Batch Operations:** Facilitates batch processing of multiple files. - **Ghidra Server Integration:** Seamless integration with Ghidra's reverse engineering capabilities. - **P-Code Emulation and Live Debugger Integration:** Enhances debugging and analysis. - **P-Code-Graph Data Flow Analysis:** Advanced data flow analysis for deeper insights. ### Tech Stack: - **Java:** The primary language used for development and integration. ### Why It Matters: The Ghidra MCP Server is essential for anyone looking to leverage the power of reverse engineering with modern AI tools and automation. Its comprehensive feature set and integration with Ghidra make it a valuable asset for developers and researchers in the field of software reverse engineering.