Interactive 3D

Embedding Space Explorer

Rotate, zoom, and hover to explore the 3D PCA projections of all four embedding spaces. Color = log(token frequency).

Embeddings

Euclidean vs Lorentz Token Embeddings

Loading Euclidean...
Loading Lorentz...

Left: Euclidean GPT-2 embeddings (PPL=48.9, ρ=+0.924). Right: Lorentz GPT-2 embeddings (PPL=43.69, ρ=−0.650). The Lorentz model places frequent tokens in a dense core (dark cluster) with rare tokens radiating outward.

Showing 1,000 stratified tokens (evenly sampled across frequency ranks) out of 16,384 for performance.

Classification Centroids

Lorentz MLR vs Poincaré HypMLR Centroids

Loading Lorentz centroids...
Loading Poincaré centroids...

Left: Lorentz GPT-2 LorentzMLR centroids (−d² logits). Right: Poincaré Hyp-Output HyperbolicMLR centroids (Ganea et al.). Both develop frequency-correlated radial structure, but the Lorentz model achieves 3.8× better PPL.