When selecting the learning rate randomly using validation data, why should the learning rate not always be set to 1?
a) A learning rate of 1 guarantees convergence.
b) Validation data is not suitable for determining the learning rate.
c) Learning rates should be adjusted based on model complexity.
d) Random selection is not a reliable method for choosing a learning rate.
