Compute & interactively visualize large embeddings Open Source • GitHub • Apple 8 150 Layout Agent 2.0 Thinks before it builds Website Builder • No-Code • Vibe coding
Embedding Atlas is a tool that provides interactive visualizations for large embeddings. It allows you to visualize, cross-filter, and search embeddings and metadata. Open sourced by Apple.
Embedding Atlas provides interactive visualization of large-scale embeddings with automatic clustering, density estimation, and real-time search capabilities, enabling users to navigate and comprehend embedding structure directly in the browser without data leaving their machine
The tool offers real-time search and nearest neighbor functionality, allowing users to query embeddings and find similar data points, supporting exploration of semantic relationships in their datasets
WebGPU implementation with WebGL 2 fallback delivers fast, smooth performance for up to several million points, with order-independent transparency ensuring accurate rendering of overlapping data
Embedding Atlas provides Python packages and Streamlit components that allow users to embed interactive visualizations directly in their existing workflows, supporting custom embedding computation and projection
Multi-coordinated views enable users to interactively link and filter data across metadata columns while visualizing embeddings, supporting exploratory analysis of relationships between embeddings and associated attributes