RAG with Open Source and Running "Examples"

Congratulations on coming this far! You've already made quite some progress. This module is to make it easy for you to build and run your applications using Examples on the LLM App repository.

It is not specific to the Basic RAG pipeline you saw earlier. But it's pretty much applicable there as well.

What are the Examples offered?

Most popular frameworks/repositories offers multiple possible use cases under its examples folder to illustrate various possible avenues of impact.

For instance, an interesting example is self-hosted real-time document AI pipelines with indexing from Google Drive/Sharepoint folders (webpage link).

For building RAG applications that leverage open source models available locally on your machine, understandably you need to refer to the "private-rag" example.

But how can you, as a developer, leverage these resources and run these examples?

Once you've cloned/forked the LLM App repository and set up the environment variables (as per the steps mentioned on this link), you're all set to run the examples.

The exact process is listed below the table which shares the types of examples you can explore. Do give it a quick read to know the possibilities of what you can build for your project. This is the complete list of examples. You can find it here.

Example Type	What It Does	What's Special	Good For
demo-question-answering	Basic end-to-end RAG app that answers queries using LLMs connected to a live data source	Supports multiple document formats (PDF, DOCX) and unstructured data.	Real-time document-based queries.
demo-document-indexing	Real-time indexing for RAG with live document updates from sources like Google Drive, SharePoint	Can be integrated with Langchain or LlamaIndex. Provides live indexing and retriever functionality to be used with any frontend	Live indexing for connected data sources
adaptive-rag	Reduces token cost in RAG by up to 4x while maintaining accuracy	Optimized token usage for lower costs	Cost-efficient RAG applications
gpt_4o_multimodal_rag	Indexes complex data (charts, tables) from unstructured documents using GPT-4o	Multimodal extraction from complex sources	Extracting from unstructured financial data
private-rag	Fully private version of the demo RAG app using local infrastructure	Keeps your data private	Sensitive environments requiring local deployment
slides_ai_search	Multi-modal search across PowerPoint and PDF slides using GPT-4o with real-time indexing and metadata extraction	Indexes locally, real-time updates, uses vision language models to parse and index slides automatically	Efficiently searching through presentations (PPTX, PDF) for sales teams or any large-scale document handling
unstructured_to_sql_on_the_fly	Converts unstructured data into SQL tables, answers queries by translating into SQL	Natural language to SQL translation with LLMs.	Advanced data manipulation and queries
drive_alert	Monitors Google Drive documents for changes and sends alerts via Slack	Real-time change monitoring	Tracking document changes and updates

Simple Way to Run the Examples on LLM App

Considering you've done the steps before, here's a recommended, step-by-step process to run the examples easily:

Considering you've done the steps before, here's a recommended, step-by-step process to run the examples easily:

1 - Open a terminal and navigate to the LLM App repository folder:

cd llm-app

2 - Choose Your Example. The examples are located in the examples folder. Say you want to run the 'drive_alert' example. You have two options here:

• Option 1: Run the centralized example runner. This allows you to quickly switch between different examples:

  python run_examples.py drive_alert

---

• Option 2: Navigate to the specific pipeline folder and run the example directly. This option is more focused and best if you know exactly which example you're interested in:

  python examples/pipelines/drive_alert/app.py

That's it! 😄

By following these steps, you're not just running code; you're actively engaging with the LLM App’s rich feature set, which can include anything from real-time data syncing to triggering alerts on critical changes in your document store.

It's a step closer to implementing your LLM application that can have a meaningful impact.