Multiple Choice

According to Machine Learning Yearning, what is the key implication if the test set is harder than the dev set when the two sets have different distributions?

Related