Llama-2 with LlamaIndex: Document Querying System

This project demonstrates how to use Meta's Llama-2 model with LlamaIndex (formerly GPT Index) for querying documents uploaded in Google Colab. The system retrieves and answers questions based on the content of PDF files you upload.

Features:

• Uses Llama-2 7B Chat model from Hugging Face.
• Integrates LlamaIndex to query PDF files.
• Runs directly in Google Colab, making it easy to set up and experiment.

---

Setup Instructions

1. Fork and Clone the Repository: Clone this repository to get started with the provided Colab notebook. You can either run this notebook on your local machine or directly in Google Colab.

   git clone <your-repo-url>

2. Upload the Notebook to Google Colab: Open Google Colab and upload the notebook file Llama2_with_llamaindex.ipynb. You can also upload it from this repo if needed.

3. Apply for Access to Llama-2: Before you can use Llama-2, ensure you have access to the model:

   • Go to the Llama-2 7B Chat Model page on Hugging Face.
   • Click Request Access and follow the steps to get permission.
   • Once approved, you’ll receive access to the model for free.

4. Generate a Hugging Face Token:

   • Visit your Hugging Face API Tokens page.
   • Create a new token and ensure you check "Read access to contents of all public gated repos you can access" in the user permissions.
   • Copy this token as you’ll need it in the next step.

5. Running the Notebook: In Google Colab:

   • Install the required packages (included in the notebook).
   • Upload any PDF files of your choice. Make sure your PDF files are stored in the /content/data/ directory.
   • The notebook uses SimpleDirectoryReader from LlamaIndex to load these PDFs:

     documents = SimpleDirectoryReader("/content/data").load_data()

   • Once the documents are loaded, you can ask any questions about the content in the query cell:

     query = "Your question about the document"
     response = query_engine.query(query)
     print(response)

6. Query Documents: After uploading the PDF files and setting up the Llama-2 model, you can start asking questions about your documents by simply running the query cell in the notebook.

---

Example Workflow

1. Upload PDF Files: Upload your files using the Colab interface.
2. Load Documents: Use the SimpleDirectoryReader to load the documents.
3. Ask a Question: Input your question in the query section, and the system will return an answer based on the document content.

---

Requirements

• Hugging Face account with access to Llama-2.
• Hugging Face API token with permission to access gated content.
• Google Colab environment.