Optimise a DSPrrr module over a grid of candidate configurations. Accepts
either an explicit grid data frame or a set of parameter definitions that
can be expanded into a grid (named lists or tidymodels parameter sets).
Usage
optimize_grid(module, ...)
# S3 method for class 'Module'
optimize_grid(
module,
data,
metric = metric_exact_match(),
grid = NULL,
parameters = NULL,
objective = c("maximize", "minimize"),
.llm = NULL,
control = list(),
...
)Arguments
- module
A DSPrrr module (created via
module()).- ...
Additional arguments forwarded to
evaluate().- data
Development data containing columns required by the module's signature plus any fields consumed by the metric.
- metric
Metric function applied per example. Defaults to
metric_exact_match(). Useas_dsprrr_metric()to adapt yardstick/vitals metrics.- grid
Optional data frame/tibble of candidate configurations.
- parameters
Optional named list or tidymodels parameter set used to generate a grid when
gridis not supplied.- objective
Optimisation direction.
"maximize"(default) selects the highest metric value;"minimize"selects the lowest.- .llm
Optional ellmer chat object reused during optimisation.
- control
Named list of control options. Recognised entries:
progress(logical),parallel(logical forwarded toevaluate()),evaluation_progress(logical),grid_type("regular"or"random"),grid_levels(integer, for regular grids), andgrid_size(integer, for random grids).