Text Embeddings Visualizer

Compute semantic embeddings for multiple texts using a neural language model, then explore their relationships as a 2D scatter plot. Texts that are semantically similar cluster together. Everything runs in your browser.

How It Works

Each text is converted into a high-dimensional numeric vector (an "embedding") using all-MiniLM-L6-v2, a lightweight sentence-transformer model running entirely in your browser via Transformers.js. The 384-dimensional embeddings are projected to 2D using Principal Component Analysis (PCA) so you can see how texts relate spatially.

Points that appear close together share similar meaning. The scatter plot uses the first two principal components, which capture the most variance in the embedding space. The similarity matrix shows pairwise cosine similarity — a value of 1.0 means identical meaning, 0 means unrelated, and negative values indicate contrasting concepts.

The model is downloaded once from a public CDN and then cached in your browser, so subsequent runs are instant. No text ever leaves your device.