Gemini + LangChain PDF QA Utility

This project shows how to use Google Gemini and LangChain to answer questions about a PDF file.

What it does

• Loads a PDF from a web link.
• Splits the PDF into chunks and creates a searchable index.
• Lets you ask questions about the PDF using Gemini.
• Shows how to generate general content and keep chat history.

Requirements

• Python 3.8 or newer
• google-generativeai
• langchain
• langchain-google-genai
• chromadb
• python-dotenv
• ipython

Install everything with:

pip install google-generativeai langchain langchain-google-genai chromadb python-dotenv ipython

or

pip install -r requirements.txt

Setup

1. Get a Gemini API key from Google AI Studio.
2. Make a .env file in this folder with:

   GEMINI_API_KEY=your_api_key_here
   

Usage

Run the script:

python gemini-utilize.py

Customization

You can change the PDF link, chunk size, or prompts in the script.

• temperature : Influences the randomness in token selection.
• top_k : Measure of how many of the most probable tokens are considered at each step.
• max_output_tokens : Upper limit of tokens generated in a response.
• top_p : Controls how the AI model chooses words when generating text.

---