Nomic-embed-text-v2

RAG & Memory 💻 Python 🔴 No recent commits
1.9k stars

Nomic Atlas Python Client is a library for interacting with Nomic Atlas, a platform for exploring, labeling, searching, and sharing large unstructured datasets. It solves the problem of analyzing and visualizing massive datasets (text, image, audio, video) by providing tools to generate embeddings, create interactive maps, and perform semantic searches. It is for developers and data scientists who need to manage and derive insights from large-scale unstructured data.

✨ Key features

  • Organize text, image, and embedding data
  • Create shareable maps with or without coding
  • Access high-level data structures and individual datapoints
  • Search millions of datapoints instantly
  • Cluster data into semantic topics
  • Tag, clean, and deduplicate datasets

🎯 Use cases

  • Visualize and explore large text corpora like news articles or social media posts
  • Build interactive maps of image or audio datasets for sharing
  • Perform semantic search over millions of embeddings
  • Automatically cluster data into topics for analysis
  • Clean and deduplicate large datasets

📦 Installation

🧰 Requirements: Requires Python, a Nomic account, and an API token obtained via nomic login.

pip install nomic

Then login:

nomic login

Follow instructions to obtain access token, then:

nomic login [token]

🚀 Usage

from nomic import atlas
import numpy as np

# Randomly generate a set of 10,000 high-dimensional embeddings
num_embeddings = 10000
embeddings = np.random.rand(num_embeddings, 256)

# Create Atlas project
dataset = atlas.map_data(embeddings=embeddings)

print(dataset)

❓ FAQ

How do I install the Nomic library?

Install via pip: pip install nomic. Then login with nomic login and follow the instructions to get your access token.

How can I access the embeddings of my dataset?

After creating a map, you can access projected and latent embeddings via dataset.maps[0].embeddings.projected and dataset.maps[0].embeddings.latent.

Can I search my data semantically?

Yes, use map.embeddings.vector_search(ids=['my_query_point'], k=5) to find nearest neighbors, then retrieve similar datapoints with dataset.get_data(ids=neighbors[0]).

What types of data does Atlas support?

Atlas supports text, image, audio, and video data, and can handle datasets from hundreds to tens of millions of points.

📊 Repository

Stars★ 1,880
Forks🍴 196
Open issues🐛 50
Last commit🕒 Nov 11, 2025
Created📅 Jul 2022
Language💻 Python

🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.