training error in cross-validation / forecast

I am using prophet in R to make a forecast for one year (360 days) with five years of available data. To evaluate the accuracy of my forecast, I aim to cross-validate it based on training data from four years (1460 days). The "cross-validation" function makes this really convenient:
`df.cv <- cross_validation(m, horizon = 365, units = 'days', initial = 1460)
df.p <- performance_metrics(df.cv)
colMeans(df.p[,3]) # average RMSE = test set error?`

With this piece of code I can easily extract the test set/validation set error (e.g., RMSE) but how do I get to the training error in prophet?