Projects

### Overview This repository provides implementable Python and Jupyter Notebook code along with benchmark datasets for learning how to classify brain signals using deep learning. It accompanies the survey on deep learning for non-invasive brain signals and the textbook "Deep Learning for EEG-based Brain-Computer Interface: Representations, Algorithms and Applications" (World Scientific, 2021). ### Key Features - Implementable Python and Jupyter Notebook code - Benchmark datasets for brain signal classification - Covers taxonomy of BCI signal paradigms by acquisition method (ECoG, EEG, fNIRS, fMRI, EOG, MEG) - Describes core deep learning architectures: MLP, RNN, LSTM, GRU, CNN, AE, VAE, GAN, DBN, GIN - Guidelines for designing BCI systems across signal categories, models, and applications - Novel BCI prototypes for authentication, visual reconstruction, language interpretation, and neurological disorder diagnosis ### Tech Stack - Python - Jupyter Notebook ### Why It Matters This repository provides practical tools and datasets for researchers and practitioners interested in developing and applying deep learning techniques for brain-computer interface applications.

Overview: This repository contains the code for a NeurIPS 2023 paper titled "Large Language Models as Commonsense Knowledge for Large-Scale Task Planning." It uses large language models as both the commonsense world model and the heuristic policy within Monte Carlo Tree Search, enabling better-reasoned decision-making for daily task planning problems. Key features: - Large Language Models (LLMs) as commonsense world model and heuristic policy within Monte Carlo Tree Search (MCTS) - Substantially reduces search complexity by providing a commonsense prior belief of states - Adapted from open-source works including data generation, testing settings, and baselines Tech stack: Python Why it matters: This approach leverages the ability of large language models to provide commonsense knowledge and heuristic guidance, significantly improving the efficiency and reasonableness of decision-making in large-scale task planning problems. Repository: 1989Ryan/llm-mcts Primary language: Python GitHub stars: 302 GitHub description: NeurIPS 2023 paper on using large language models as commonsense knowledge for large-scale task planning. GitHub repository for the project.

### Overview: GPT-voice-conversation-chatbot (formerly GPT3-STTC) is a conversational AI bot that allows you to have an engaging and safely emotive spoken conversation with ChatGPT or GPT-4 using your microphone. It remembers what you've said for the session and can develop a memory of you over time if desired. ### Key Features: - Engaging and emotive spoken conversations - Terminal-based interaction - Memory of previous conversations - GPT-4 API integration ### Tech Stack: Python ### Why It Matters: This tool provides a unique way to interact with ChatGPT through voice, making the conversation more interactive and personal. It's particularly useful for those who prefer to speak rather than type, and it can remember previous interactions, enhancing the conversation experience. ### Repository: Adri6336/gpt-voice-conversation-chatbot - GitHub Stars: 302 - GitHub Description: Allows you to have an engaging and safely emotive spoken / CLI conversation with the AI ChatGPT / GPT-4 while giving you the option to let it remember things discussed. - Primary Language: Python - GitHub Status: https://status.openai.com/ - Releases: Stable, but not all features are available. Cloning the repository and running 'git pull' periodically will get the newest features and bug fixes. ### Installing: - **Windows:** 1. Download Python from https://www.python.org/ 2. Download the repo via releases, git cloning, or code button

### Overview: Video generation models have demonstrated significant improvements in speed and efficiency, enabling faster and more accurate video processing. This project, "Less is Enough: Training-Free Video Diffusion Acceleration via Runtime-Adaptive Caching," addresses the challenges of video synchronization and performance bottlenecks by introducing a novel method called "EasyCache." This technique allows video generation models to run significantly faster without the need for extensive training, making it particularly useful for real-time applications and low-latency scenarios. ### Key features: - **Runtime-Adaptive Caching**: A method to cache intermediate results of video generation models, reducing the need for retraining and speeding up the inference process. - **Network Load Mitigation**: Automatically adjusts the cache to handle varying network conditions, ensuring consistent performance across different environments. - **Compatibility with SVG**: EasyCache enhances the compatibility of video generation models with SVG, improving the visual output quality. ### Tech stack: - Python ### Why it matters: This project is crucial for developers and researchers working on video processing and generation, as it significantly speeds up the inference process, making it more efficient and practical for real-time applications and low-latency environments.

### Overview: Welcome to ML-ProjectYard! This is a collection of multiple machine learning-based projects, including frontend and backend, designed to help developers master ML and AI algorithms. Projects are deployable and beginner-friendly, making it an ideal resource for both beginners and experienced developers. ### Key Features: - **Projects Deployable Ready:** All projects are ready for deployment, making them easy to use and integrate. - **Beginner-Friendly:** Suitable for both beginners and experienced developers, providing a solid foundation in ML and AI. - **Projects Cover:** Includes projects in Machine Learning, Deep Learning, Natural Language Processing, Computer Vision, and more. ### Tech Stack: - **Jupyter Notebook:** Used for data analysis and project development. ### Why It Matters: - **Mastering ML/AI Algorithms:** These projects will help you master various machine learning and AI algorithms, making you a pro in the field. - **Versatile Projects:** From predicting Airbnb prices to building a web app for checking banknote authenticity, these projects cover a wide range of applications, ensuring you can apply your skills in various contexts. ### GitHub Stars: - **Stars:** 301 - **Description:** This is the one-stop open-source destination for all the projects based on Machine Learning, Deep Learning, NLP, Computer Vision, and everything includes frontend and backend! ### Repository: - **Repository:** ashishsahu1/ML-ProjectYard - **Primary Language:** Jupyter Notebook - **GitHub Stars:** 301 ### Contributing: - **Contrib

Overview: This project is a scikit-learn wrapper designed to fine-tune Google's BERT model for various text and token sequence tasks. It includes configurable MLP as a final classifier/regressor for text and text pair tasks, token sequence classifier for NER, POS, and chunking tasks, and pretrained models for scientific and biomedical domains. Key features: - Configurable MLP as final classifier/regressor for text and text pair tasks - Token sequence classifier for NER, POS, and chunking tasks - Pretrained models for scientific and biomedical domains (SciBERT, BioBERT) Tech stack: Python 3.5 and PyTorch 0.4.1 Why it matters: This project provides a user-friendly interface for fine-tuning BERT models, making it accessible for developers and researchers to leverage BERT's capabilities for natural language processing tasks without the need for extensive knowledge of deep learning frameworks.

### Overview: Loomic is an open-source AI canvas design agent that allows users to communicate directly with AI to generate images and videos on an infinite canvas. It eliminates the need for timeframes, templates, or learning prompt engineering, making it a flexible and data-driven alternative to Lovart, CapCut Video Studio, and Canva AI. ### Key Features: - **Conversation-Based Design**: Users can chat with the AI to generate, edit, and arrange designs on an unbounded canvas. - **Model Integration**: Supports multiple models including Google Imagen, Gemini Image, Vertex AI, OpenAI DALL-E 3, Replicate Flux Kontext, SDXL, Recraft, Seedream, and more. - **Video Generation**: Generates videos using Google Veo 3.1, 3.0, and 2.0, and Replicate Kling, Seedance, Wan, Sora, and Hailuo. - **Canvas-Based Interface**: Utilizes Excalidraw for a flexible and interactive design experience. - **Brand Kit**: Allows users to set brand colors, fonts, and logos, ensuring consistency. - **Integration with LemonSqueezy**: Provides a built-in integration for LemonSqueezy subscriptions. - **Infinite Canvas**: Supports multiple layers, resizing, and layering of designs. - **Integration with Supabase**: Stores and manages data, including integration with Google Fonts. ### Tech Stack: - **Frontend**: TypeScript, React, Tailwind CSS, LangGraph, Fastify, Supabase, Excalidraw, Pnpm

### Overview: embodied agents is a toolkit designed to seamlessly integrate state-of-the-art transformer models into robotics stacks. It provides a consistent, reliable, scalable, and customizable solution that can be easily integrated into existing robot environments. ### Key Features: - **Transformer Integration**: Seamlessly integrates transformer models into robotics stacks. - **Consistency and Reliability**: Ensures that the integration is consistent and reliable across different environments and setups. - **Scalability**: Easily scalable to handle various robot and sensor configurations. - **Configurability**: Highly configurable to match any observation and action space. ### Tech Stack: Python ### Why It Matters: embodied agents simplifies the integration of advanced AI models into robotics, making it easier and more efficient to develop and deploy intelligent robots.

Overview: This curated list is a collection of resources aimed at machine learning engineers, covering essential skills and information for developing and implementing ML solutions. Key features: - Communication: Resources for improving communication skills. - Software Engineering: Tools and techniques for API design and software engineering best practices. - Machine Learning: Articles and guides on machine learning and its applications. - DevOps: Resources for DevOps practices and methodologies. Tech stack: Python, Markdown, HTML, CSS Why it matters: This list provides a comprehensive overview of essential resources for machine learning engineers, covering communication, software engineering, machine learning, and DevOps practices.

Overview: Textpipe is a Python package designed to clean raw text and extract metadata such as word count, sentence count, and named entities from text documents. Key features: - Cleans raw text by removing HTML and other unreadable constructs - Identifies the language of text - Extracts the number of words, number of sentences, and named entities from a text - Calculates the complexity of a text - Provides sentiment analysis (polarity and subjectivity score) - Generates word counts - Computes minhash for cheap similarity estimation of documents Tech stack: Python Why it matters: Textpipe simplifies the process of extracting metadata and cleaning text data, making it useful in various applications such as content management, text analysis, and data preprocessing for machine learning models. Repository: textpipe/textpipe Primary language: Python GitHub stars: 302

### Overview: Engrim simplifies the complex world of AI coding by providing a unified, local-first memory engine that seamlessly integrates with various AI models and environments, ensuring that your project's decisions and state are preserved across different tools and platforms. ### Key Features: - **Cross-Model Integration**: Engrim allows developers to switch between Google Antigravity, Claude Code, Cursor, Codex CLI, GitHub Copilot CLI, and OpenCode on the same codebase without losing architectural decisions or project state. - **Episodic Memory**: A local-first SQLite file stores high-precision, curated episodic working memory, ensuring precision and continuity of context. - **Zero Cloud Lock-In**: The engine is designed to avoid cloud-based storage, keeping your project's decisions and state local and private. ### Tech Stack: Python ### Why It Matters: Engrim eliminates the "200,000-token context window trap," ensuring that your coding models remain sharp and focused, while preserving your project's decisions and state across different environments and tools. This makes it an essential tool for developers looking to streamline their workflow and maintain consistency in their projects.

Overview: Vaporetto is a fast and lightweight pointwise prediction-based tokenizer designed for efficient text processing tasks. Key features: - Supports both distribution models and KyTea models for tokenization. - Provides APIs for Rust and CLI interfaces. - Compatible with various models and formats, including zstd and KyTea. Tech stack: Rust Why it matters: Vaporetto offers a highly optimized solution for tokenizing text efficiently, making it suitable for applications requiring rapid response times and high throughput.

### Overview: The NLP Newsletter is a weekly or biweekly newsletter that provides quick summaries of the latest natural language processing (NLP) and machine learning (ML) stories across various categories such as fairness & ethics, society, educational resources, and publications. The newsletter aims to highlight important developments in the field and encourage contributions from the community. ### Key Features: - **Weekly or Biweekly Summaries**: Provides concise summaries of recent NLP and ML news. - **Categories Covered**: Includes topics like fairness & ethics, society, educational resources, and publications. - **Community Engagement**: Encourages contributions through translations and translations of past issues. ### Tech Stack: The newsletter is written in Markdown and is hosted on GitHub Pages. ### Why It Matters: The NLP Newsletter serves as a valuable resource for researchers, practitioners, and enthusiasts interested in the latest advancements in NLP and ML. By providing quick summaries and insights, it helps keep the community informed and engaged with the evolving landscape of these fields. ### Repository: The repository is maintained by the Dair AI organization and can be found at [dair-ai/nlp_newsletter](https://github.com/dair-ai/nlp_newsletter). ### GitHub Stars: The repository has received 301 stars, indicating its popularity and active community engagement. ### How to Contribute: - **Project Page**: Visit the project page for ongoing tasks and issues related to the research project. - **Issues**: Check the issues tab for good first issues and translations. - **Slack Group**: Join the Slack group for asking questions and

Overview: This repository contains codes and corpora for the paper "Effective Neural Solution for Multi-Criteria Word Segmentation" (accepted & forthcoming at SCI-2018). It provides a simple solution for multi-criteria Chinese word segmentation using Python. Key features: - Multi-criteria Chinese word segmentation - Codes and corpora for the paper - Dependency: Python3, dynet - Quick Start: Prepare corpora, split into train/dev/test sets, convert to pickle files, and perform training and testing. Tech stack: Python3, dynet Why it matters: This project offers a practical solution for Chinese word segmentation, which is crucial for natural language processing tasks.

Overview: chatless 是一款基于 Tauri 2.0 和 Next.js 的本地优先 AI 聊天客户端，支持接入多种云端 AI 服务和本地模型，提供文档解析、图片分析、本地 RAG 知识库、MCP 协议集成、Prompt 管理和 WebDAV 同步功能，保护用户隐私。 Key features: - 支持多种 AI 提供商 - 文档解析与本地 RAG 知识库 - 图片分析与本地模型 - MCP 协议集成 - Prompt 管理 - WebDAV 同步功能 Tech stack: Tauri 2.0、Next.js 15、TypeScript、Rust、SQLite Why it matters: 保护用户隐私，提供本地存储和本地处理能力，支持多种 AI 服务和本地模型，增强用户体验和数据安全性。 Repository: kamjin3086/chatless Primary language: TypeScript GitHub stars: 301 GitHub description: 💻一款简洁实用轻量级的本地AI对话客户端，采用Tauri2.0和Next.js编写 A simple, practical, and lightweight local AI chat client, written in Tauri 2.0 & Next.js. README excerpt: <div align="center"> <img src="public/logo.svg" alt="chatless logo" /> </div> <h1 align="center">chatless</h1> <p align="center"> 基于

Overview: ark-nlp 是一个开源的自然语言处理库，旨在快速实现当前最先进的 NLP 模型解决方案。 Key features: - 预训练模型：包括 BERT、ERNIE1.0、NEZHA、Roformer 和 ERNIE-CTM。 - 文本分类：使用经典 RNN、CNN、GRU/LSTM 结构或预训练模型进行文本分类。 - 文本匹配：基于 BERT/ERNIE 的无监督 Simcse 算法。 - 命名实体识别：支持 CRF BERT、Biaffine BERT、Span BERT、Global Pointer BERT 和 Efficient Global Pointer BERT。 - 关系抽取：采用统一的词-词关系分类方法进行关系抽取。 Tech stack: Python Why it matters: ark-nlp 提供了全面的 NLP 工具包，简化了模型复现和开发过程，使得研究人员和开发者能够快速构建和评估 NLP 模型。

Overview: This project is a Rust-based MCP server designed to fetch and provide accurate documentation for specific Rust crates, ensuring that AI coding assistants have up-to-date and contextually relevant information. Key features: - Focuses on a single Rust crate per server instance. - Allows specifying required crate features for documentation generation. - Uses OpenAI's `text-embedding-3-small` model for semantic search. - Leverages OpenAI's `gpt-4o-mini-2024-07-18` model for LLM summarization. - Caches generated documentation content and embeddings for speed. - Runs as a standard MCP server over stdio, integrating with the LLM assistant. Tech stack: Rust, OpenAI APIs. Why it matters: This server addresses the issue of outdated code suggestions by providing a focused, up-to-date knowledge source for specific Rust crates, ensuring that AI coding assistants can generate more accurate and relevant code suggestions. Repository: Govcraft/rust-docs-mcp-server GitHub stars: 299

### Overview: This repository introduces the official ComfyUI native node supporting **InfiniteYou** with FLUX, providing a robust framework for flexible and high-fidelity identity-preserved image generation. ### Key Features: - **InfiniteYou (InfU)**: A framework that leverages Diffusion Transformers (DiTs) for image generation while preserving identity. - **Identity Injection**: InfuseNet injects identity features into the DiT base model via residual connections, enhancing identity similarity. - **Text-Image Alignment**: Improves text-image alignment, image quality, and alleviates face copy-pasting issues. - **Pretraining and SFT**: Includes pretraining and supervised fine-tuning with synthetic single-person-multiple-sample (SPMS) data for better performance. ### Tech Stack: Python ### Why It Matters: This framework is a significant contribution to the field of image generation, offering a flexible and high-quality solution for identity-preserved image creation while addressing common issues in existing methods.

Overview: AUITestAgent is a natural language-driven GUI testing tool designed for mobile apps, capable of fully automating the entire process of GUI interaction and function verification. Key features: - Automatically generates and conducts UI interactions based on natural language test requirements. - Performs verification after the interaction. - Supports multiple languages and platforms. Tech stack: Python, Selenium, Natural Language Processing (NLP). Why it matters: AUITestAgent simplifies the GUI testing process, making it more accessible and efficient for developers, especially those working with complex mobile applications.

Overview: This repository provides a comprehensive workshop guide on how to apply artificial intelligence (AI) in ethical hacking, covering various applications such as AI reconnaissance, intelligent screenshot analysis, smart content discovery, automated exploit generation, and more. Key features: - AI Reconnaissance - Intelligent Screenshot Analysis - Smart Content Discovery - Automated Exploit Generation - Hackbots - MCP Integrations Tech stack: Python Why it matters: This repository offers practical insights into how AI can be effectively utilized in offensive security, making it a valuable resource for both learning and practical application in ethical hacking. Repository: ethiack/ai4eh GitHub stars: 298 GitHub description: AI for Ethical Hacking - Workshop