gaussian.experiment

What is trained in the Gaussian study, and early validation of the training settings.

Classes

GaussianExperiment

What is trained: problem, prior, training data, model and training procedure.

Module Contents

class gaussian.experiment.GaussianExperiment

Bases: nnpd.Experiment

What is trained: problem, prior, training data, model and training procedure.

Selected by “experiment” in the training settings. The runner owns execution and files.

name = 'gaussian-prior-dependence'

version = '1'

make_problem(settings)

make_prior(settings, member, problem)

members(settings)

A fixed comparison cohort within each sweep configuration.

sample_training(context, writer)

build_model(context)

train(context, model)

Modify model in place; return JSON-serializable training history.

signature(stage, context)

Override for cross-run sharing; conservative default uses all training settings.

sources(stage)

Include modules implementing transitive scientific dependencies.

validate(settings)

Fail before any sampling or training, including for invalid alternatives.

estimate(settings, member)

Training workload shown by the plan.