Week1_Assignment3_Jazz_Improvization sampling

When We sampling novel sequences for W1_assignment 3 Jazz Improvisation, why are we using tf.math.argmax to select the value with the highest possibility?

The reason I ask the aforementioned question is because for W1_assignment 2 Dinosaur Island, when we sample novel sequences to generate new dinosaur name, we did not choose the value with the highest possibility. Instead, we used np.random.choice to sample randomly.

The two assignments use different methods. I think it is because grading is easier if the results aren’t random.