# apple/embedding-atlas

Embedding Atlas is a tool that provides interactive visualizations for large embeddings. It allows you to visualize, cross-filter, and search embeddings and metadata.

Repository: https://github.com/apple/embedding-atlas
Canonical: https://ross.abutalabs.com/products/embedding-atlas
Homepage: https://apple.github.io/embedding-atlas/
Language: TypeScript
License: MIT
License Family: permissive
Topics: embedding, visualization
Last push: 2026-08-26T04:33:47+00:00

## Health v2 (maintenance only)
Score: 86/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 98, longevity 34
- inputs: {"age_days": 484, "days_push": 7, "days_rel": 14, "gap_med": 20, "n_releases_24m": 22}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4921, forks 321 (observed 2026-08-28T04:09:02.454143+00:00)

## What it is
Embedding Atlas is an open-source tool from Apple for interactive visualization of large embeddings and their metadata, supporting clustering, density contours, real-time nearest-neighbor search, and cross-filterable dashboards rendered with WebGPU. It ships as a Python CLI and Jupyter widget plus npm components for React and Svelte, and exposes an MCP server so AI agents can query and chart the data.

## Use cases
- visualize a few million embeddings interactively in the browser
- explore clusters and outliers in embedding datasets with density contours
- find nearest neighbors for a query embedding in real time
- build cross-filtered dashboards over tabular metadata alongside embeddings
- inspect text, image, and audio data points inside an embedding projection
- let AI agents explore and chart a dataset via MCP
- embed an embedding visualization widget in a Jupyter notebook

## When to choose
- you need to explore or debug large embedding collections (e.g., from LLM or ML pipelines)
- you want interactive clustering, search, and cross-filtering without building custom tooling
- you need a notebook widget or reusable React/Svelte components for embedding views
- your datasets scale to millions of points and need GPU-accelerated rendering

## When to avoid
- you need general-purpose BI dashboards without embeddings
- you require a production analytics database or long-term data storage rather than an exploration tool
- your environment lacks WebGPU-capable browsers for large-scale rendering

## Facets
- artifact type: application
- maturity: active
- function: data-visualization, search-engine, machine-learning, mcp, cli
- domain: data-visualization, machine-learning, large-language-models, data-science, developer-tools
- platform: cli, python, cross-platform
- tags: embeddings, webgpu, nearest-neighbors, clustering, jupyter-widget, cross-filtering, react-components, svelte-components, web-server, nodejs

## Member repositories
- apple/embedding-atlas (main) score 86

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:02.454143+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:18:01.956803+00:00, confidence not recorded.
  - readme: https://github.com/apple/embedding-atlas (fetched 2026-08-28T04:09:02.454143+00:00, sha 15943c67ef97)
  - homepage: https://apple.github.io/embedding-atlas/ (fetched 2026-08-29T09:00:13.400818+00:00, sha f089b2e56bda)
  - registry_npm: https://registry.npmjs.org/embedding-atlas (fetched 2026-08-29T09:00:13.409820+00:00, sha 85d13bd14431)
- Data as of 2026-08-30T08:39:29.467469+00:00.