Chat with books using DeepInfra and LlamaIndex

Published on 2024.06.07 by Oguz Vuruskaner

Chat with books using DeepInfra and LlamaIndex header picture

As DeepInfra, we are excited to announce our integration with LlamaIndex. LlamaIndex is a powerful library that allows you to index and search documents using various language models and embeddings. In this blog post, we will show you how to chat with books using DeepInfra and LlamaIndex.

We will be using the Project Gutenberg library to get the text of the book "Crime and Punishment" by Fyodor Dostoevsky. We will then use the Meta Llama 3 70B language model and the MiniLM embedding model to chat with the book.


  • Python 3.9 or higher
  • DeepInfra API Key


First, let's create a virtual environment and activate it:

python3 -m venv venv
source venv/bin/activate

Here are the required packages to install:


Let's install them:

pip install llama-index llama-index-llms-deepinfra llama-index-embeddings-deepinfra

Before getting started, we also need to get the API key for DeepInfra. You can get your DeepInfra API key from here.

Let's create a .env file in the root directory of the project and add the following lines:


Code Implementation

Here's a Python script to chat with the book "Crime and Punishment":

import requests
from dotenv import load_dotenv, find_dotenv
import re

_ = load_dotenv(find_dotenv())

from llama_index.core import VectorStoreIndex, Document

from llama_index.llms.deepinfra import DeepInfraLLM
from llama_index.embeddings.deepinfra import DeepInfraEmbeddingModel

LLM = "meta-llama/Meta-Llama-3-70B-Instruct"
EMBEDDING = "sentence-transformers/all-MiniLM-L12-v2"
BOOK_TITLE = "Crime and Punishment"

def maybe_get_gutenberg_book_id(title):
    url = f"{title}"
    response = requests.get(url)
    books = response.json()["results"]
    for book in books:
        if title.lower() in book["title"].lower():
            return book["id"]
    return None

def get_document(book_id):
    url = f"{book_id}/{book_id}-0.txt"
    response = requests.get(url)
    text = response.text
    # Get rid of binary characters.
    text = re.sub(r"[^\x00-\x7F]+", "", text)
    return Document(text=text)

if __name__ == "__main__":

    llm = DeepInfraLLM(LLM, max_tokens=1000)
    embed_model = DeepInfraEmbeddingModel(EMBEDDING)

    book_id = maybe_get_gutenberg_book_id(BOOK_TITLE)
    document = get_document(book_id)

    index = VectorStoreIndex.from_documents([document], embed_model=embed_model)
    chat_engine = index.as_chat_engine(
        llm=llm, embed_model=embed_model, max_iterations=20

    response =
        "Summarize the discussion between Raskolnikov and Pyotr Petrovich"

    # The conversation between Raskolnikov and Pyotr Petrovich takes place at the office of...


Voila! You have successfully chatted with the book "Crime and Punishment" using DeepInfra and LlamaIndex. You can now use this code snippet to chat with any book of your choice. Enjoy reading!

For more information on LlamaIndex, please visit our LLM documentation and Embedding documentation.

Feel free to experiment with other books and questions to explore the capabilities of DeepInfra. See you in the next blog post!

Happy chatting! 📚🦙