Smoke render: optimisation and MCMC are shrunk for CI speed, so figures and diagnostics are not publication-fidelity. The production docs are built with full-length inference.

GPJax

Real#

class gpjax.parameters.Real(value)[source]#

Bases: AbstractUnwrappable

Unconstrained parameter. unwrap() returns the value unchanged.

Expand for references to gpjax.parameters.Real

Backend Module Design

Dual Parameterisation of Sparse GPs (t-SVGP)

Natural Gradients in Practice

Natural Gradients

Parameters:: value (Array)

unwrap()[source]#: Returns the unwrapped pytree, assuming no wrapped subnodes exist.