RAG开源工具 是 AI Skill Hub 本期精选AI工具之一。综合评分 7.5 分，整体质量较高。我们推荐使用将其纳入你的 AI 工具库，帮助提升工作效率。
**为什么要使用开源工具而非商业 SaaS？**
对于个人开发者和有隐私需求的用户，本地部署的开源工具意味着数据不离本机，不受第三方服务商的数据政策约束。同时，开源工具通常没有使用次数限制和月度费用，一次安装即可长期使用，对于高频使用场景的总拥有成本（TCO）远低于订阅制商业工具。
**安装与环境准备**
RAG开源工具 依赖 Python 运行环境。建议通过 pyenv（Python）或 nvm（Node.js）管理 Python 版本，避免全局环境污染。对于新手用户，推荐先创建虚拟环境（python -m venv venv && source venv/bin/activate），再安装依赖，这样即使出现问题也可以随时删除虚拟环境重新开始，不影响系统稳定性。
**社区与维护**
GitHub Issue 和 Discussion 是获取帮助的最快渠道。在提问前建议先检查 Closed Issues（已关闭的问题），大多数常见问题都已有解答。遇到 Bug 时，提供 pip list 的输出、完整错误堆栈和最小可复现示例，能显著提高开发者响应速度。AI Skill Hub 将持续追踪 RAG开源工具 的版本更新，及时通知重要功能变化。
基于Python的RAG系统，支持多模块和多语言，开源AI工具，值得关注。
RAG开源工具 是一款基于 Python 开发的开源工具，专注于 RAG、Gemma3、Python 等核心功能。作为 GitHub 开源项目，它拥有活跃的社区支持和持续的版本迭代，代码完全透明可审计，支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流，都能提供稳定可靠的解决方案。
基于Python的RAG系统，支持多模块和多语言，开源AI工具，值得关注。
RAG开源工具 是一款基于 Python 开发的开源工具，专注于 RAG、Gemma3、Python 等核心功能。作为 GitHub 开源项目，它拥有活跃的社区支持和持续的版本迭代，代码完全透明可审计，支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流，都能提供稳定可靠的解决方案。
- 开源免费，支持本地部署，数据完全自主可控
- 活跃的 GitHub 开源社区，持续迭代更新
- 提供详细文档和使用示例，新手友好
- 支持自定义配置，灵活适配不同使用环境
- 可作为基础组件集成进现有技术栈或进行二次开发
- 本地部署运行，保护数据隐私，满足合规要求
- 自定义集成到现有系统，扩展技术栈能力
- 作为开源基础组件进行商业化二次开发
# 方式一：pip 安装（推荐）
pip install rag-with-gemma3
# 方式二：虚拟环境安装（推荐生产环境）
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install rag-with-gemma3
# 方式三：从源码安装（获取最新功能）
git clone https://github.com/Bbs1412/rag-with-gemma3
cd rag-with-gemma3
pip install -e .
# 验证安装
python -c "import rag_with_gemma3; print('安装成功')"
- 访问 GitHub 仓库页面
- 按照 README 文档完成依赖安装
- 根据系统环境完成初始化配置
- 参考官方示例或文档开始使用
- 遇到问题可在 GitHub Issues 中查找解答
# 命令行使用
rag-with-gemma3 --help
# 基本用法
rag-with-gemma3 input_file -o output_file
# Python 代码中调用
import rag_with_gemma3
# 示例
result = rag_with_gemma3.process("input")
print(result)
# rag-with-gemma3 配置文件示例（config.yml） app: name: "rag-with-gemma3" debug: false log_level: "INFO" # 运行时指定配置文件 rag-with-gemma3 --config config.yml # 或通过环境变量配置 export RAG_WITH_GEMMA3_API_KEY="your-key" export RAG_WITH_GEMMA3_OUTPUT_DIR="./output"
`RAG with Gemma-3`
This project is a modular Retrieval-Augmented Generation (RAG) system built with Google DeepMind's - Gemma 3 served locally using Ollama. It allows users to upload documents (PDF, TXT, Markdown etc.), and then chat with the content using natural language queries - all processed through a local setup for privacy and full control.
Designed with modularity and performance in mind, the system handles end-to-end workflows including file ingestion, vector embedding, history summarization, document retrieval, context-aware response generation, and streaming replies to a frontend. It supports multi-file embeddings per user, persistent session history and document storage, and offers live document previews - making it complete end-to-end RAG pipeline useful for educational, or personal assistants.
Check out the live project deployment: 
Features
- User Authentication: + Authenticate users using SQLite-3 database and bcrypt based password hashing and salt. 
+ Store user data securely and also auto clear stale sessions data.
- UI and User Controls: + Build a responsive UI using Streamlit app. Provide a chat interface for users to ask questions about their documents, get file previews and receive context-aware responses. 
+ User uploaded files and corresponding data are tracked in a SQLite-3 database. + Allow users to delete their uploaded documents, and manage their session history. 
+ Note: Files previews are cached for 10 minutes, so even after deletion, the file preview might be available for that duration. + Also works with FastAPI SSE to show real-time responses from the LLM and retrieved documents and metadata for verification. 
+ UI supports thinking models also to show the LLM's thought process while generating responses. 
- User wise document management:
- Support multi-file embeddings per user, allowing users to upload multiple documents and retrieve relevant information based on their queries.
- Some documents can also be added as public documents, which can be accessed by all users. (like shared rulebooks or manuals or documentation)
- Embeddings, Vector Storage and Retrieval:
- Implement vector embeddings using
LangChaincomponents to convert documents into vector representations. - Open source
mxbai-embed-largemodel is used for generating embeddings, which is a lightweight and efficient embedding model. - Use
FAISSfor efficient vector storage and retrieval of user-specific + public documents. - Integrate similarity search and document retrieval with Gemma-based LLM responses.
- FastAPI Backend: + Build a FastAPI backend to handle file uploads, document processing, user authentication, and streaming LLM responses. + Integrate with 'LLM System' module to handle LLM tasks. + Provide status updates to UI for long running tasks: 
+ Implement Server-Sent Events (SSE) for real-time streaming of LLM responses to the frontend with NDJSON format for data transfer. 
+ Provide UI with retrieved documents and metadata for verification of responses.
- LLM System: + Modular LLM System using LangChain components for: 1. Document Ingestion: Load files and process them into document chunks. 1. Vector Embedding: Convert documents into vector representations. 1. History Summarization: Summarize user session history for querying vector embeddings and retrieving relevant documents. 1. Document Retrieval: Fetch relevant documents based on standalone query and user's metadata filters. 1. History Management: Maintain session history for context-aware interactions. 1. Response Generation: Generate context-aware responses using the LLM. 1. Tracing: Enable tracing of LLM interactions using LangSmith for debugging and monitoring LLM interactions. 1. Models: Use Ollama to run the Gemma-3 LLM and mxbai embeddings locally for inference, ensuring low latency and privacy.
- Dockerization:
- Create a dynamic
Dockersetup for easy deployment as either a development or deployment environment. - Use
Dockerfileto manage both FastAPI and Streamlit server in a single container (mainly due to Hugging Face Spaces limitations).
🛠️ Installation
There are two ways to run this project - either directly using a Virtual Environment or using Dockerfile.
🐋 Docker
Dockerfile is coded dynamically to support both development and deployment environments.
Docker Files (Why 4?)
There are two Docker methods in this repo (single-container vs multi-container), and that is why you see four Docker-related files.
1. Single-container (Hugging Face / simple local run) - Dockerfile: Builds one container that runs FastAPI + Streamlit together via entrypoint.sh. - Use this when: + You are deploying on Hugging Face Spaces (no Compose support), or + You want a single container locally with ENV_TYPE=dev or ENV_TYPE=deploy. - Uses root requirements.txt for a full install.
1. Multi-container (Local Docker Compose) - Dockerfile.fastapi: Backend-only container, tuned for local dev (Ollama on host). - Dockerfile.streamlit: Frontend-only container, Streamlit UI. - docker-compose.yml: Wires them together for local dev. - Uses docker/requirements_be.txt and docker/requirements_fe.txt to avoid bloat. - Update .streamlit/secrets.toml when using Compose: + Set server.ip_address = "http://fastapi:8000" (container-to-container URL, not localhost).
TL;DR: UseDockerfilefor HF (single container). Usedocker-compose.ymlfor local dev (two containers).
RAG Samples:
- Q: Highest possible grade: 
- Q: Formatted Output: 
Virtual Environment
1. Clone the repository:
git clone --depth 1 https://github.com/Bbs1412/rag-with-gemma3.git
1. Create virtual environment and install dependencies:
# Create environment:
python -m venv venv
# Activate environment:
source venv/bin/activate # On Linux/Mac
# or
venv\Scripts\activate # On Windows
# Install dependencies:
pip install -r requirements.txt
1. (Optional) If you want to use LangSmith tracing, create a .env file in the server directory and add these credentials:
# ./server/.env
LANGCHAIN_TRACING_V2=true
LANGCHAIN_ENDPOINT="https://api.smith.langchain.com"
LANGCHAIN_API_KEY="<paste_your_api_key_here>"
LANGCHAIN_PROJECT="rag-with-gemma3"
1. Start the FastAPI server:
cd server
uvicorn server:app
# For development with hot-reloading
# uvicorn server:app --reload --port 8000
1. Start the Streamlit server:
cd ..
streamlit run app.py
1. You can now access these servers: - FastAPI backend at http://localhost:8000 - Streamlit frontend at http://localhost:8501 - FastAPI Swagger UI at http://localhost:8000/docs
Environment Configurations:
1. Development: - Project uses http://host.docker.internal:11434 as the Ollama server for local inference. - This is to ensure that existing Ollama models in the host machine are accessible from the docker container. - In this env, all three ports {8000:FastAPI, 8501:Streamlit, 11434:Ollama} are exposed for easy access. 1. Deployment: - Project uses Google Gemini-2.0-Flash-Lite as the LLM and text-embedding-004 as embedding model. - Primarily due to deployment and API limitations of Gemma3 model. - In this env, only port 7860 is exposed for the Streamlit frontend.
#### Development: 1. Build the Docker image:
docker build -t BBS/rag-with-gemma3:dev --build-arg ENV_TYPE=dev .
1. Create a Docker container:
docker create --name rag-gemma-cont-dev \
-e ENV_TYPE=dev \
# Below 4 are optional env-vars for LangSmith tracing \
-e LANGCHAIN_TRACING_V2=true \
-e LANGCHAIN_ENDPOINT="https://api.smith.langchain.com" \
-e LANGCHAIN_API_KEY="<paste_your_api_key_here>" \
-e LANGCHAIN_PROJECT="rag-with-gemma3" \
# Port mapping for FastAPI, Streamlit, and Ollama \
-p 8000:8000 -p 8501:8501 -p 11434:11434 \
BBS/rag-with-gemma3:dev
1. Start the Docker container:
docker start -a rag-gemma-cont-dev
1. You can now access these servers: - FastAPI backend at http://localhost:8000 - Streamlit frontend at http://localhost:8501
#### Deployment: 1. Build the Docker image:
docker build -t BBS/rag-with-gemma3:prod --build-arg ENV_TYPE=deploy .
1. Create a Docker container:
docker create --name rag-gemma-cont-prod \
-e ENV_TYPE=deploy \
# Below 4 are optional env-vars for LangSmith tracing \
-e LANGCHAIN_TRACING_V2=true \
-e LANGCHAIN_ENDPOINT="https://api.smith.langchain.com" \
-e LANGCHAIN_API_KEY="<paste_your_api_key_here>" \
-e LANGCHAIN_PROJECT=deployed-rag-gemma3 \
# This is necessary env variable \
-e GOOGLE_API_KEY="<paste_your_google_api_key_here>" \
# Port mapping, only 7860 is exposed \
-p 7860:7860 \
BBS/rag-with-gemma3:prod
1. Start the Docker container:
docker start -a rag-gemma-cont-prod
- You can now access the Project at http://localhost:7860
To test some sub-components:
- This ensures that relative imports work correctly in the project.
cd server
python -m llm_system.utils.loader
该工具基于Python，支持多模块和多语言，开源AI工具，值得关注，但需要进一步优化和完善。
- 开源免费，支持本地部署，数据完全自主可控
- 活跃的 GitHub 开源社区，持续迭代更新
- 提供详细文档和使用示例，新手友好
- 支持自定义配置，灵活适配不同使用环境
- 可作为基础组件集成进现有技术栈或进行二次开发
- 本地部署运行，保护数据隐私，满足合规要求
- 自定义集成到现有系统，扩展技术栈能力
- 作为开源基础组件进行商业化二次开发
- +GPL-3.0 协议，可免费商用
- +完全开源免费，无授权费用
- +本地部署，数据完全自主可控
- +开发者社区支持，遇问题可查可问
- −安装和初始配置可能需要一定技术基础
- −功能完整性通常不如成熟商业产品
- −技术支持主要依赖开源社区，响应速度不稳定
AI Skill Hub 为第三方内容聚合平台，本页面信息基于公开数据整理，不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后，再部署至生产环境，并做好必要的安全评估。
⚠️ GPL 3.0 — 强 Copyleft，衍生作品须开源，含专利保护条款，不可闭源使用。
经综合评估，RAG开源工具 在AI工具赛道中表现稳健，质量良好。如果你已有明确的使用需求，可以直接上手体验；如果还在评估阶段，建议对比同类工具后再做决策。
| 原始名称 | rag-with-gemma3 |
| 原始描述 | 开源AI工具：This project is a modular Retrieval-Augmented Generation (RAG) system built with。⭐12 · Python |
| Topics | RAGGemma3Python |
| GitHub | https://github.com/Bbs1412/rag-with-gemma3 |
| License | GPL-3.0 |
| 语言 | Python |
收录时间：2026-05-23 · 更新时间：2026-05-23 · License：GPL-3.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。