Random Search vs Grid Search

Medium

machine-learning Google Meta DeepMind

You're tuning a model with 6 hyperparameters, each with a wide range of plausible values, and you have a fixed compute budget of 100 training runs. Why would random search typically outperform grid search here, given the same budget?

Multiple Choice

Random search always finds the global optimum, while grid search only finds local optima With a fixed budget, random search samples more distinct values per hyperparameter, which matters when only a few dimensions actually affect performance Grid search cannot be parallelized across multiple machines, while random search can Random search requires less memory than grid search for the same number of runs

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

Random Search vs Grid Search

Multiple Choice

Related problems

Deep-dive guides