MongoDB/mongodb-docs-chunked

Overview

This dataset consists of chunked and embedded versions of a small subset of MongoDB's technical documentation. It contains 795 chunks from 498 documentation pages, sampled from 50 documentation sources.

Dataset Structure

The dataset consists of the following fields:

• chunk_id: Unique identifier for the chunk.
• page_id: Identifier of the documentation page the chunk came from.
• source_name: The documentation source, such as pymongo or atlas-cli.
• product_name: The product the page documents, when available.
• url: Link to the documentation page.
• title: Title of the documentation page.
• text: Content of the chunk in Markdown format.
• embedding: The embedding of the chunk's content, created using Voyage AI's voyage-4-large model with 1024 dimensions and input_type="document".

Pages were split into chunks of up to ~2,000 tokens at Markdown heading boundaries. Because the Voyage 4 models share an embedding space, you can embed queries for this dataset with voyage-4-large, voyage-4, or voyage-4-lite.

Usage

This dataset can be useful for prototyping RAG applications. It is derived from mongodb-eai/docs.

Ingest Data

To experiment with this dataset using MongoDB Atlas, first create a MongoDB Atlas account.

You can then use the following script to load this dataset into your MongoDB Atlas cluster:

import os

from datasets import load_dataset
from pymongo import MongoClient

client = MongoClient(os.environ["MONGODB_URI"])
collection = client["your_database_name"]["mongodb_docs"]

dataset = load_dataset("MongoDB/mongodb-docs-chunked", split="train")
collection.insert_many(dataset.to_list())

print(f"Ingested {len(dataset)} documents")