HAL56

Llama Index Gradio Interface

This project provides a simple Gradio interface for using the Llama Index, a versatile library for semantic search. This interface allows you to interact with the Llama Index through a chat-like interface using the Gradio library.

Getting Started

Prerequisites

• Make sure you have Python installed (version 3.6 or above).

Installation

1. Clone the repository:

   git clone https://github.com/Saideep-23/HAL56.git
   cd HAL56

pip install -r requirements.txt mkdir sample_data

Place your sample documents in the sample_data directory

python your_script_name.py

Code Explanation

The provided code sets up a Gradio interface for interacting with the Llama Index. Here's a brief explanation of the key components:

Llama Index Initialization:

• The code initializes the Llama Index using the VectorStoreIndex, SimpleDirectoryReader, and ServiceContext classes.
• It loads sample data from the sample_data directory.

LlamaCPP Configuration:

• Configures the LlamaCPP model with various parameters such as temperature, max_new_tokens, and more.

Service Context Setup:

• Creates a ServiceContext with default settings, specifying the chunk size, LlamaCPP model, and embedding model.

Rerank Model Setup:

• Initializes a SentenceTransformerRerank model for reranking search results.

Query Engine Setup:

• Configures the query engine with similarity settings and node postprocessors, including the previously defined rerank model.

Prediction Function:

• Defines a prediction function that queries the Llama Index and returns the response.

Gradio Interface Launch:

• Uses Gradio to launch a chat interface for the prediction function.