minRAG - A minimalist and powerful RAG system designed for ease of use and flexibility.

## Overview of minRAG minRAG is a Retrieval-augmented Generation (RAG) system designed to enhance text generation quality and accuracy by integrating language models with information retrieval techniques. It is built with a focus on simplicity and power, featuring a codebase of no more than 10,000 lines, no installation requirements, and the ability to launch with a double-click. ## Supported AI Platforms minRAG supports multiple AI platforms, including OpenAI, Gitee AI, Baidu Qianfan, Tencent Cloud LKE, Alibaba Cloud Bai Lian, and ByteDance Volcano Engine. This flexibility allows users to integrate minRAG into various environments. ## Overview of minRAG minRAG improves text generation by combining language models with information retrieval techniques. This approach retrieves relevant information and augments it with language model outputs, enhancing the quality and accuracy of the generated text. ## Key Features of minRAG The key features of minRAG include: - Extreme simplicity and power, making it user-friendly and efficient. - A codebase limited to no more than 10,000 lines, ensuring maintainability and readability. - No installation required, allowing users to start by double-clicking. - Support for multiple AI platforms, providing flexibility across different environments. ## Usage and Accessibility of minRAG Users can access and use minRAG by simply double-clicking to launch the system, as it requires no installation. This ease of use lowers the technical barrier, making it suitable for both technical and non-technical users. ## Primary Source for minRAG The primary source for minRAG information is its GitHub repository, accessible at [https://github.com/minrag/minRAG](https://github.com/minrag/minRAG). While the user-provided URL [https://code.exmay.com](https://code.exmay.com) was mentioned, research indicates that GitHub is the main source for accurate and up-to-date information. ## Community Recognition of minRAG minRAG was recognized as the "Best Popular Project" in the 2021 OSC China Open Source Project Selection, highlighting its community recognition and utility. ## Components of minRAG minRAG uses components such as FTS5 for BM25 full-text search, Vec for vector search, and implements various modules like MarkdownConverter, DocumentSplitter, OpenAIDocumentEmbedder, SQLiteVecDocumentStore, OpenAITextEmbedder, VecEmbeddingRetriever, FtsKeywordRetriever, DocumentChunksReranker, PromptBuilder, OpenAIChatMessageMemory, OpenAIChatCompletion, and Pipeline. These components support pipeline settings and extensions. ## Supported AI Platforms The default AI platform for minRAG is Gitee AI, which offers 100 free calls per day. Users can configure other AI platforms as needed. ## Language Support in minRAG minRAG supports both Chinese (zh-CN) and English (en-US). Language files are located in the `minragdatadir/locales` directory and can be configured via `minragdatadir/install_config.json`. ### Citation sources: - [minRAG](https://github.com/minrag/minRAG) - Official URL Updated: 2025-03-31