This application demonstrates the implementation of Retrieval Augmented Generation (RAG) and a RAG Agent using the LangChain and LangGraph libraries. The application loads documents from arXiv, creates a vector store using FAISS, and builds a retrieval-augmented generation chain. It also creates a RAG Agent that can perform a series of actions to answer questions based on the retrieved context.

• Loads documents from arXiv related to "Retrieval Augmented Generation"
• Splits the documents into chunks using RecursiveCharacterTextSplitter
• Creates a FAISS vector store from the chunked documents
• Builds a retrieval-augmented generation chain using the vector store and OpenAI's GPT-3.5-turbo model
• Implements a RAG Agent with a tool belt containing DuckDuckGoSearchRun and ArxivQueryRun
• Creates a workflow for the RAG Agent using LangGraph's StateGraph
• Integrates the retrieval-augmented generation chain into the RAG Agent's workflow
• Determines if a question is fully answered by the response using GPT-4
• Invokes the RAG Agent to answer questions based on the retrieved context

Before running the application, make sure you have the following:

• Python 3.7 or higher
• OpenAI API key (set as an environment variable named OPENAI_API_KEY)
• Required Python packages (listed in the requirements.txt file)

1. Clone the repository:

git clone https://github.com/your-username/your-repository.git

2. Navigate to the project directory:

cd your-repository

3. Install the required packages:

pip install -r requirements.txt

4. Set your OpenAI API key as an environment variable:

export OPENAI_API_KEY=your-api-key

To run the application, execute the following command:

python main.py

The application will load the documents, create the vector store, build the retrieval-augmented generation chain, and create the RAG Agent. It will then invoke the RAG chain and the RAG Agent with sample questions and print the responses.

The code is organized into the following main components:

• RetrievalAugmentedGeneration: Handles loading documents, creating the vector store, and building the retrieval-augmented generation chain.
• RAGAgent: Implements the RAG Agent with a tool belt, creates the workflow using LangGraph's StateGraph, and integrates the retrieval-augmented generation chain into the workflow.
• main: The entry point of the application, where the components are instantiated, and sample questions are asked.

The application relies on the following main dependencies:

• LangChain: A framework for developing applications with large language models.
• LangGraph: A library for building and executing workflows using state graphs.
• FAISS: A library for efficient similarity search and clustering of dense vectors.
• OpenAI: The OpenAI API for accessing language models like GPT-3.5-turbo and GPT-4.

• The LangChain and LangGraph communities for providing powerful tools and resources.
• OpenAI for their advanced language models and APIs.
• The authors of the research papers and documents used in this application.