wrong penalty checking in get_glmnet_coefs()

[EmilHvitfeldt]

When trying to rerender https://www.tidymodels.org/learn/models/coefficients/ in tidymodels/tidymodels.org#92, i found the following bug in get_glmnet_coefs(). Below is a reprex

library(tidymodels)
data(Chicago)

Chicago <- Chicago %>% select(ridership, Clark_Lake, Austin, Harlem)

glmnet_spec <- 
  linear_reg(penalty = 0.1, mixture = 0.95) %>% 
  set_engine("glmnet")

glmnet_wflow <- 
  workflow() %>% 
  add_model(glmnet_spec) %>% 
  add_formula(ridership ~ .)

glmnet_fit <- fit(glmnet_wflow, Chicago)

tidy(glmnet_fit, penalty = 5.5620)
#> 
#> Attaching package: 'Matrix'
#> The following objects are masked from 'package:tidyr':
#> 
#>     expand, pack, unpack
#> Loaded glmnet 4.1-8
#> Error in `tidy()`:
#> ! `penalty` must be a number between 0 and 1 or `NULL`, not the number 5.56.

^{Created on 2025-03-07 with reprex v2.1.1}

[blackrow]

Yes, I am encoutring the same issue today while solving a problem. Imposing a penalty between 0 and 1 is not natural at all. Is there any workaround ?

[topepo]

That is a bug induced when we updated our error checkers. I'll put in a PR immediately.