A hatchery incubator-temperature trainer shows a loss that explodes after a few steps. Someone's first fix is to run more epochs. What should change first?
Select an answer to reveal the explanation.
Short Explanation
Think of incubator-temperature loss exploding after a few steps. Lower the learning rate. More epochs will not pull a diverging job back down.
Full Explanation
An overly large learning rate is a common cause of a diverging, exploding loss. Adjust that hyperparameter before adding epochs. Extra epochs and a still-larger rate worsen the jump. Transcribe is not a trainer control.