Skip to main content

User guide

Retrieval-Augmented Generation (RAG) allows you to ask questions against your own documents using advanced language models. This guide walks you through each step of the workflow, from deploying the model to retrieving precise, document-based answers.

Steps to Use the RAG Feature​

1. Deploy a Model​

  • Start by choosing a suitable model for your application from the Hugging Face selection available through the Qubrid platform.
  • Deploy the model directly from the platform. You may be able to select models based on language tasks or performance needs.

2. Open the RAG UI​

  • Once your model is deployed, head over to Actions → RAG UI in the Qubrid interface.
  • This opens the Retrieval-Augmented interface, where you’ll interact with your documents.

3. Upload Document(s)​

  • Use the upload option to add one or more PDF files (or other supported formats) to the platform.
  • The system supports batch uploads, allowing you to query across multiple documents at once.

4. Wait for Processing​

  • After uploading, the system automatically parses and indexes your document(s).
  • This usually takes just a few seconds, regardless of document size, preparing them for rapid search and retrieval.

5. Start Chatting​

  • With indexing complete, you can now ask questions using the chat interface provided.
  • The model retrieves relevant information from your uploaded documents and generates precise, well-formatted answers.
  • This allows for efficient knowledge extraction tailored directly to your data.

Additional Tips​

  • Supported File Types: Always check which formats (PDF, TXT, etc.) are supported for upload.
  • Handling Large Documents: For very large files, allow a brief extra moment for processing and indexing.
  • Effective Questioning: The more specific your query, the higher the accuracy and detail of the returned answers.
  • Multiple Documents: You may ask questions referencing multiple files—the system automatically selects relevant content.

By following these steps, you can efficiently leverage the RAG feature to extract, analyze, and interact with the knowledge contained in your documents using state-of-the-art AI models.