RAG AI Search

An intelligent search system based on RAG (Retrieval-Augmented Generation) that supports multiple LLM models, including GPT and local Ollama models.

For Chinese version, please check README_cn.md

Features

• Support for multiple LLM models (GPT and local Ollama models)
• Streaming output generation
• Display of reference document sources
• History record saving and querying
• Responsive interface design

To Do

• Optimize real-time search
• Optimize generation result efficiency, such as first detecting whether to search, then searching, otherwise directly generating
• Support for multi-round dialogue

System Requirements

• Python 3.8+
• Flask
• Azure OpenAI API access (for GPT)
• Ollama service (for local models)

Installation Steps

1. Clone the project

git clone https://github.com/your-repo/rag-ai-search.git

2. Install dependencies

pip install -r requirements.txt

3. Install Ollama

• Install Ollama

  Reference: Ollama Installation

• Install Ollama model after installation

  ollama run llama3:8b

• Configure Ollama service address

  Configure Ollama service address in config/settings.py

  OllamaConfig.BASE_URL = "http://127.0.0.1:4008"

4. Install Search Engine

• Install search engine

  Reference: Install Search Engine

• Configure search engine

  Configure search engine in config/settings.py

  SearchConfig.SEARX_BASE_URL = "http://127.0.0.1:11434"

5. Configure Azure OpenAI API

   Create a .env file in the project root directory and configure the following environment variables:

gpt4o_mini_api_key='your_api_key'
gpt4o_mini_api_version='your_api_version'
gpt4o_mini_azure_endpoint='your_azure_endpoint'

GPT calling method: utils/gpt4_client.py

Note: If you don't have a GPT key, you can directly use the Ollama model without configuring Azure OpenAI

Starting the Search System

1. Start Web Service

python web/app.py

2. Open browser and visit: http://localhost:5000

3. Select Model:

   • GPT: Uses Azure OpenAI service
   • Ollama Model: Uses locally deployed model

4. Enter your question and wait for the answer

   • System will display reference document sources
   • Real-time streaming display of generated results
   • Support for viewing history records

Technical Architecture