Error when examining aggregated SHAP values in R using tidymodels

I have trained and fitted a C5.0 model with adaptive boosting to predict a binary outcome. This was performed using tidymodels and the parsnip package.

I am now trying to evaluate the model in more detail using DALEX and DALEXtra.

The outcome variable is "yes/no" so I have converted these to binary "1/0". I have created an explainer using the following code with the DALEXtra package:

explain_c50 <- 
  explain_tidymodels(model = c5_final_fit, 
                     data = final_data_test, 
                     y = y_test, 
                     verbose = F)

I have also created an explainer using only the DALEX package, and encounter the exact same issue:

custom_predict <- function(object, newdata) {
  pred <- 
    predict(object, newdata, type = 'prob')[1] %>% 
    pull(.pred_F)
  
  return(pred)
}


DALEX_explainerTest <- DALEX::explain(model = c5_final_fit, 
                                  data = final_data_test,
                                  predict_function = custom_predict,
                                  y = y_test, 
                                  label = "c50-train")

These run without error. However, when I try to run a command to interrogate additive SHAP values I get the following error:

DALEX::shap_aggregated(explainer = explain_c50, new_observations = final_data_test[1:10, ])

Error message is always the same, no matter what I try:

Error in `[<-.data.frame`(`*tmp*`, , candidate, value = list(pure_model_prediction = list( : 
  replacement element 1 is a matrix/data frame of 1 row, need 21

I have tried processing the data slightly differently, tried using the training data (rather than test), etc. The error message is always this way so I assume I am making a fairly fundamental error. I cannot understand why I am getting this error.

Does anyone reading this have any idea?