tuning_view

Back to L4.5 | Browse all axes | Browse all options

Axis tuning_view on sub-layer L4_5_D_tuning_history (layer l4_5).

Sub-layer

L4_5_D_tuning_history

Axis metadata

  • Default: 'multi'

  • Sweepable: False

  • Status: operational

Operational status summary

  • Operational: 4 option(s)

  • Future: 0 option(s)

Options

cv_score_distribution – operational

Distribution of CV scores at each iteration.

L4.5.D tuning view cv_score_distribution.

This option configures the tuning_view axis on the L4_5_D_tuning_history sub-layer of L4.5; output is emitted under manifest.diagnostics/l4_5/L4_5_D_tuning_history/ alongside the other selected views.

When to use

Detecting high-variance objective surfaces; wide distributions suggest the search has not converged.

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: objective_trace, hyperparameter_path, multi

Last reviewed 2026-05-05 by macroforecast author.

hyperparameter_path – operational

Sequence of hyperparameter values explored.

L4.5.D tuning view hyperparameter_path.

This option configures the tuning_view axis on the L4_5_D_tuning_history sub-layer of L4.5; output is emitted under manifest.diagnostics/l4_5/L4_5_D_tuning_history/ alongside the other selected views.

When to use

Diagnosing search behaviour – e.g. detecting Bayesian optimisation getting stuck on a local minimum.

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: objective_trace, cv_score_distribution, multi

Last reviewed 2026-05-05 by macroforecast author.

multi – operational

Produce all tuning-history views together.

L4.5.D tuning view multi.

This option configures the tuning_view axis on the L4_5_D_tuning_history sub-layer of L4.5; output is emitted under manifest.diagnostics/l4_5/L4_5_D_tuning_history/ alongside the other selected views.

When to use

Comprehensive tuning audit. Activates the multi branch on L4.5.tuning_view; combine with related options on the same sub-layer for a comprehensive diagnostic.

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: objective_trace, hyperparameter_path, cv_score_distribution

Last reviewed 2026-05-05 by macroforecast author.

objective_trace – operational

Tuning-objective trace over iterations.

L4.5.D tuning view objective_trace.

This option configures the tuning_view axis on the L4_5_D_tuning_history sub-layer of L4.5; output is emitted under manifest.diagnostics/l4_5/L4_5_D_tuning_history/ alongside the other selected views.

When to use

Default convergence audit; monotone decrease confirms good search behaviour.

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: hyperparameter_path, cv_score_distribution, multi

Last reviewed 2026-05-05 by macroforecast author.