Hyperparameter Tuning: Bayesian vs Grid Search

Medium

machine-learning Google Databricks Meta

You need to tune 6 hyperparameters for a model that takes 10 minutes to train. Grid search with 5 values per parameter would require 5^6 = 15,625 evaluations. What is a more efficient alternative?

Multiple Choice

Random search, which samples randomly but always requires the same number of evaluations as grid search Bayesian optimization, which uses past evaluations to guide the search toward promising regions Manual tuning, which is always faster than automated methods Reduce the number of hyperparameters to 2, since tuning more is never practical

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

Hyperparameter Tuning: Bayesian vs Grid Search

Multiple Choice

Related problems

Deep-dive guides