Skip to content

VicoErv/normal-scoring-playground

Repository files navigation

Normal Scoring Playground

Most people learn the Normal distribution as a "bell curve." This playground ignores the curve and focuses entirely on the Score—the vector field that tells a point where to go.

Key Features

  • No Inertia: Experience first-order dynamics where $v = \text{score}(x)$, leading to perfect, non-oscillatory convergence.
  • Anisotropy: Warp the space with Covariance (Sigma) and Rotation to see how paths bend.
  • Nonlinear Residuals: Layer cubic terms over the linear Gaussian core to see how "real-world" scoring deviates from the ideal.
  • Physics Comparison: Contrast score-based transport with robotic constant-speed motion and second-order spring physics.

About

An interactive 2D playground for building intuition about Gaussian Scoring as a vector field and first-order data transport.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors