AI Bootcamp in India - Homework 3

Overview

This project implements a RAG (Retrieval-Augmented Generation) application using the Google Gemini API. It was originally designed for OpenAI's API but has been adapted due to token limitations and system resource constraints.

Background

Initially, this project was set up to use OpenAI's API. However, we encountered several challenges:

1. OpenAI API token limitations (5000 tokens) were insufficient for this project.
2. Creating the output.json file was time-consuming and caused system overheating.
3. System errors occurred twice during processing.

To address these issues, we switched to the Google Gemini API, which offers a free tier with higher token limits.

Prerequisites

• Python 3.10+
• Google Gemini API Key

Setup

1. Clone this repository:

   git clone <repository-url>
   cd <repository-name>

2. Install dependencies:

   pip install -r requirements.txt

3. Create a .env file in the project root and add your Google Gemini API key:

   GOOGLE_API_KEY=your_google_gemini_api_key_here

Known Issues and Troubleshooting

1. Deprecated Imports: You may see warnings about deprecated imports from langchain. To resolve this, update the imports in rag_implementation.py:

   from langchain_community.vectorstores import Chroma
   from langchain_community.embeddings import GooglePalmEmbeddings
   from langchain_community.llms import GooglePalm

2. API Key Not Found: If you encounter an error like "Did not find google_api_key", ensure that:

   • Your .env file is in the correct location (project root).
   • The API key in the .env file is correct and properly formatted.
   • You're running the script from the project root directory.

3. Transcript Issues: The transcript of the one-hour video may sometimes cause issues. If you encounter problems related to the transcript, try:

   • Checking the docs/intro-to-llms-karpathy.txt file for any formatting issues.
   • Splitting the transcript into smaller chunks if it's too large.

Running the Project

1. Implement the RAG system:

   python rag_implementation.py

2. Generate answers to the question list:

   python generate_answers.py

3. Evaluate the results:

   python eval.py path/to/your/output.json

Troubleshooting

If you encounter the error "TypeError: str expected, not NoneType" when setting the API key, try modifying the rag_implementation.py file: