During a data analysis task, the team decides to conduct a hyperparameter search to tune a neural network model on an NVIDIA GPU cluster.
Which two of the following practices will help maximize the efficiency of the hyperparameter tuning process? (Select two)
A. Perform the search sequentially on a single GP
B. Use only default hyperparameters as a baseline for all models.
C. Leverage NVIDIA's automatic mixed precision (AMP) during tuning.
D. Prioritize hyperparameter tuning on CPU nodes to free up GPUs for other tasks.
E. Utilize distributed hyperparameter tuning across multiple GPUs.