74 lines
2.3 KiB
Markdown
74 lines
2.3 KiB
Markdown
|
# Document-Based Question Answering with OpenWebUI
|
||
|
|
|
||
|
|
This document explains how to use the document-based question answering (RAG) functionality in the AI service.
|
||
|
|
|
||
|
|
## Overview
|
||
|
|
|
||
|
|
The AI service now supports Retrieval Augmented Generation (RAG) by leveraging OpenWebUI's built-in knowledge database. This allows users to:
|
||
|
|
|
||
|
|
1. Upload documents to OpenWebUI
|
||
|
|
2. Ask questions about those documents
|
||
|
|
3. Receive responses that incorporate information from the documents
|
||
|
|
|
||
|
|
## How It Works
|
||
|
|
|
||
|
|
When RAG is enabled:
|
||
|
|
|
||
|
|
1. The AI service forwards the request to OpenWebUI with `use_knowledge=True`
|
||
|
|
2. OpenWebUI searches its knowledge database for relevant information
|
||
|
|
3. The retrieved information is used to augment the model's response
|
||
|
|
4. The response is returned to the user
|
||
|
|
|
||
|
|
## Using RAG in API Requests
|
||
|
|
|
||
|
|
To enable RAG in your API requests, set the `use_rag` parameter to `true`:
|
||
|
|
|
||
|
|
```json
|
||
|
|
POST /chats/{chat_id}/messages
|
||
|
|
{
|
||
|
|
"message": "What information do you have about project X?",
|
||
|
|
"user_id": "user123",
|
||
|
|
"use_rag": true
|
||
|
|
}
|
||
|
|
```
|
||
|
|
|
||
|
|
## Testing RAG Functionality
|
||
|
|
|
||
|
|
You can test the RAG functionality using the `/test-rag` endpoint:
|
||
|
|
|
||
|
|
```
|
||
|
|
POST /test-rag?query=What information do you have about project X?
|
||
|
|
```
|
||
|
|
|
||
|
|
This will return a response that includes information from documents in OpenWebUI's knowledge database.
|
||
|
|
|
||
|
|
## Uploading Documents to OpenWebUI
|
||
|
|
|
||
|
|
To use RAG effectively, you need to upload documents to OpenWebUI:
|
||
|
|
|
||
|
|
1. Log in to OpenWebUI at your configured URL (default: http://104.225.217.215:8080)
|
||
|
|
2. Navigate to the Knowledge section
|
||
|
|
3. Upload your documents (PDF, TXT, DOCX, etc.)
|
||
|
|
4. OpenWebUI will automatically process and index the documents
|
||
|
|
|
||
|
|
## Troubleshooting
|
||
|
|
|
||
|
|
If RAG is not working as expected:
|
||
|
|
|
||
|
|
1. Ensure OpenWebUI is running and accessible
|
||
|
|
2. Check that documents are properly uploaded and indexed in OpenWebUI
|
||
|
|
3. Verify that the `use_rag` parameter is set to `true` in your requests
|
||
|
|
4. Check the logs for any errors related to OpenWebUI API calls
|
||
|
|
|
||
|
|
If there are connection issues with OpenWebUI, the AI service will automatically fall back to using the direct Ollama API without RAG.
|
||
|
|
|
||
|
|
## Configuration
|
||
|
|
|
||
|
|
The following configuration settings affect RAG functionality:
|
||
|
|
|
||
|
|
- `OPENWEBUI_URL`: URL of your OpenWebUI instance
|
||
|
|
- `OPENWEBUI_API_KEY`: API key for OpenWebUI (if required)
|
||
|
|
- `API_TIMEOUT`: Timeout for API requests (in seconds)
|
||
|
|
|
||
|
|
These can be set in your environment variables or in the `.env` file.
|