Error with relaxed poisson regression in tidymodels

Question

I have been trying to fit a relaxed regularized poisson regression using tidymodels and poissonreg. Whenever I specify the family (either family = poisson() or family = "poisson") as arguments to pass to the engine (glmnet), I get this error:

Error in glmnet.path(x, y, weights, lambda, nlambda, lambda.min.ratio, : Invalid family argument; must be either character, function or family object

Without the relax = TRUE argument, there is no error. Am I missing something?

Ultimately, I'd like to estimate a family = MASS::negative.binomial(theta) regression using the tidymodels framework.

library(tidymodels)
library(poissonreg)

# Data generation ----
set.seed(123)
df <- tibble(
  y = rpois(1000, lambda = 3),
  x_1 = 2 * y + rnorm(1000),
  x_2 = 0.1 * y + rnorm(1000)
)

# this works
poisson_reg(penalty = 0.5) %>%
  set_engine("glmnet",
             relax = TRUE
  ) |> 
  fit(y~., data = df) 
#> parsnip model object
#> 
#> 
#> Call:  glmnet::glmnet(x = maybe_matrix(x), y = y, family = "poisson",      relax = ~TRUE) 
#> Relaxed
#> 
#>    Df  %Dev %Dev R  Lambda
#> 1   0  0.00   0.00 1.63700
#> 2   1 14.05  77.14 1.49200
#> 3   1 25.46  77.14 1.35900
#> 4   1 34.75  77.14 1.23800
#> 5   1 42.33  77.14 1.12800
#> 6   1 48.51  77.14 1.02800
#> 7   1 53.58  77.14 0.93690
#> 8   1 57.73  77.14 0.85360
#> 9   1 61.14  77.14 0.77780
#> 10  1 63.94  77.14 0.70870
#> 11  1 66.25  77.14 0.64570
#> 12  1 68.14  77.14 0.58840
#> 13  1 69.70  77.14 0.53610
#> 14  1 70.99  77.14 0.48850
#> 15  1 72.06  77.14 0.44510
#> 16  1 72.93  77.14 0.40550
#> 17  1 73.66  77.14 0.36950
#> 18  1 74.26  77.14 0.33670
#> 19  1 74.76  77.14 0.30680
#> 20  1 75.17  77.14 0.27950
#> 21  1 75.50  77.14 0.25470
#> 22  1 75.79  77.14 0.23210
#> 23  1 76.02  77.14 0.21150
#> 24  1 76.21  77.14 0.19270
#> 25  1 76.37  77.14 0.17560
#> 26  1 76.50  77.14 0.16000
#> 27  1 76.61  77.14 0.14570
#> 28  1 76.70  77.14 0.13280
#> 29  1 76.78  77.14 0.12100
#> 30  1 76.84  77.14 0.11030
#> 31  1 76.89  77.14 0.10050
#> 32  1 76.94  77.14 0.09153
#> 33  1 76.97  77.14 0.08340
#> 34  1 77.00  77.14 0.07599
#> 35  1 77.02  77.14 0.06924
#> 36  1 77.04  77.14 0.06309
#> 37  1 77.06  77.14 0.05749
#> 38  1 77.08  77.14 0.05238
#> 39  1 77.09  77.14 0.04773
#> 40  1 77.10  77.14 0.04349
#> 41  1 77.10  77.14 0.03962
#> 42  1 77.11  77.14 0.03610
#> 43  1 77.12  77.14 0.03290
#> 44  1 77.12  77.14 0.02997
#> 45  1 77.13  77.14 0.02731
#> 46  1 77.13  77.14 0.02488
#> 47  1 77.13  77.14 0.02267
#> 48  1 77.13  77.14 0.02066
#> 49  1 77.14  77.14 0.01882
#> 50  1 77.14  77.14 0.01715
#> 51  1 77.14  77.14 0.01563

# this fails
poisson_reg(penalty = 0.5) %>%
  set_engine("glmnet",
             family = "poisson",
             relax = TRUE
  ) |> 
  fit(y~., data = df)
#> Error in glmnet.path(x, y, weights, lambda, nlambda, lambda.min.ratio, : Invalid family argument; must be either character, function or family object

^{Created on 2024-02-02 with reprex v2.0.2}