Axiom Choice Latent Space Explorer

Initializing Latent Space...

Loading mathematical modules...

Semantic token maps to ℝ³ via 512-dimensional embedding projection

Latent θ

0.000

rad

Latent φ

0.000

rad

Radius r

1.000

units

Velocity

0.000

u/s

Shannon Entropy

0.000

bits

Lyapunov Exponent

0.000

Fractal Dimension

0.000

Gaussian Curvature

0.000

Show Trajectory Traces Show Manifold Surface Show Poincaré Section Quality: Medium

Poincaré Section (y = 0)

Phase space intersection reveals attractor structure

This visualization implements:

Johnson-Lindenstrauss Embedding: 512→3 dimensional projection preserving pairwise distances
Attention Mechanism: Scaled dot-product attention from transformer architecture
Gradient Flow: SGD with momentum on loss manifold
Chaos Analysis: Real-time Lyapunov exponent calculation

60 FPS 0 particles 0ms 0 points