Is this training error, rather than validation (or test) set error? Because the graphs exhibit very little evidence of overfitting, even when model complexity is maximized according to either hyperparameter.
Iโm wondering if Iโm actually managing to tune hyperparameters using validation data, or whether the optimization is carried out using only the training data (thereby trivializing the hyperparameter search). Any insights would be highly appreciated. Thank you in advance for any advice!
