Comparing model performance shows the Error in h(simpleError(msg, call))

Question

I'm performing ridge and lasso regression for feature selection. My actual issue is related to the comparison between the two models, for example:

models <- list(ridge = ridge, lasso = lasso)

If I check the models output:

$ridge

Call:  cv.glmnet(x = x_train, y = y_train, type.measure = "mse", alpha = 0,      family = "binomial") 

Measure: Mean-Squared Error 

    Lambda Index Measure      SE Nonzero
min  3.518    51  0.2656 0.02807      86
1se  5.103    47  0.2916 0.02601      86

$lasso

Call:  cv.glmnet(x = x_train, y = y_train, type.measure = "mse", alpha = 1,      family = "binomial") 

Measure: Mean-Squared Error 

      Lambda Index Measure      SE Nonzero
min 0.003518    51 0.04640 0.02799      16
1se 0.020603    32 0.07214 0.03088      17

I know that the best model is the one that one that minimizes the prediction error, so I use the following line:

resamples(models) %>% 
  summary(metric = "RMSE")

That line must give me a summary and a comparison between the two models. However, I got the following error:

Error in h(simpleError(msg, call)) : 
  error in evaluating the argument 'object' in selecting a method for function 'summary': values must be length 1,
 but FUN(X[[1]]) result is length 0

So far I can not understand how to fix this.

This option only works with the test data or can I use the complete set too?

What could be the potential issue here? or explanation?