Chat with your PDF files for free, using Langchain, Groq, Chroma vector store, and Jina AI embeddings. This repository contains a simple Python implementation of the RAG (Retrieval-Augmented-Generation) system. The RAG model is used to retrieve relevant chunks of the user PDF file based on user queries and provide informative responses.
Follow these steps:
- Clone the repository
git clone https://github.com/S4mpl3r/chat-with-pdf.git - Create a virtual environment and activate it (optional, but highly recommended).
python -m venv .venv Windows: .venv\Scripts\activate Linux: source .venv/bin/activate - Install required packages:
python -m pip install -r requirements.txt - Create a .env file in the root of the project and populate it with the following keys. You'll need to obtain your api keys:
JINA_API_KEY=<YOUR KEY> GROQ_API_KEY=<YOUR KEY> HF_TOKEN=<YOUR TOKEN> HF_HOME=<PATH TO STORE HUGGINGFACE MODEL> - Run the program:
python main.py
You can customize the behavior of the system by modifying the constants and parameters in the main.py file:
- EMBED_MODEL_NAME: Specify the name of the Jina embedding model to be used.
- LLM_NAME: Specify the name of the language model (Refer to Groq for the list of available models).
- LLM_TEMPERATURE: Set the temperature parameter for the language model.
- CHUNK_SIZE: Specify the maximum chunk size allowed by the embedding model.
- DOCUMENT_DIR: Specify the directory where PDF documents are stored.
- VECTOR_STORE_DIR: Specify the directory where vector embeddings are stored.
- COLLECTION_NAME: Specify the name of the collection for the chroma vector store.
Kudos to the amazing libraries and services listed below:
MIT