Check your understanding of this module before moving on.
?
0/3
Checkpoint quiz 3 questions
A quick check on this module's key ideas before you move on
Question 01
What typically happens if the learning rate is set far too high?
B is correct. A learning rate that's too large causes each weight update to jump past the minimum of the loss surface, and successive overshoots can compound, making the loss bounce around or blow up rather than settle down.
Question 02
Why might a training schedule decay the learning rate over time, rather than keeping it constant?
A is correct. Early in training, large steps help cover a lot of ground quickly. As training approaches a good region of the loss surface, a smaller learning rate avoids overshooting and lets the optimizer fine-tune the weights.
Question 03
What is the idea behind a 'warmup' period at the start of training?
A is correct. Right after random initialization, large updates can be especially destabilizing since the model's internal representations are not yet meaningful. Warmup ramps the learning rate up gradually, giving the model a chance to settle before taking full-sized steps.