pre-compute metrics_info(metrics)

[simonpcouch]

The metrics_info helper is a small helper to reformat as_tibble.metric_set for easier filtering/selecting. It's reasonably fast, but was called in a couple places inside the tune_grid_loop_iter interior loops, so ends up being called thousands of times even for small grid searches. We can thus pre-compute that output and pass it as an argument to tune_grid_loop_iter. Note that this introduces merge conflicts with #641—glad to take care of that once we get there.

With main dev:

library(tidymodels)

set.seed(1)

bench::mark(
  total = fit_resamples(linear_reg(), mpg ~ ., bootstraps(mtcars, 100)),
  iterations = 5
)
#> # A tibble: 1 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 total         2.71s    2.73s     0.363    46.3MB     6.96

...with a 4% decrease in evaluation time with this PR:

#> # A tibble: 1 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 total          2.6s    2.62s     0.377    45.3MB     7.23

That effect is more substantial when tuning hyperparameters. :)

^{Created on 2023-03-17 with reprex v2.0.2}

[EmilHvitfeldt]

Amazing! Looking good

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.