How to update/finalize tabnet model with selected best parameters?

Hi all,

I'm following the tabnet tutorial RStudio AI Blog: torch, tidymodels, and high-energy physics
The author has shown how easy it is to tune the hyperparameter with the tidymodels framework but did not finalize the model with the best tuned parameters. I encountered error message when I finalize the workflow

best <- res %>%
  select_best(metric = "rmse")

final_wf <- wf %>% 

Error in update.default(object = list(args = list(epochs = ~10, penalty = ~1e-06,  : 
  need an object with call component

Is it possible to update tabnet model with workflow or am I missing something here?

Hi @tellyshia ,

Thanks for reporting!

This was a bug and should be fixed in the current dev version of tabnet.

Thanks for looking into this @dfalbel