Embedding constellation

Embed your sentences and explore their relationships in a three-dimensional, orbitable space.

How this experiment works

MiniLM computes 384-dimensional embeddings. Principal component analysis projects them to three axes for inspection.

What it does not prove

PCA discards dimensions. Distances in the view can distort semantic relationships; consult the original cosine scores.

Try this challenge

Add a sentence about an unrelated subject. Does its location agree with its cosine scores?

Related Trust Agent courses · Sources and licences

Enable JavaScript to run the experiment on your device.