Deep Learning
Using a learning rate that’s too low will cause the training process to be very slow.
True
Proposed by @svpino
Card 1 of 259. Answer: True
True
Proposed by @svpino
Card 1 of 259. Answer: True
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
True. A learning rate that is too low makes gradient descent take tiny steps, so the loss falls slowly and training needs far more iterations to converge. The learning rate η scales every update: w ← w − η ∇L(w).
Example: take L(w) = w², whose gradient is 2w, starting at w = 1. Each step multiplies w by (1 − 2η). With η = 0.1 the factor is 0.8, and w falls below 0.01 after 21 steps. With η = 0.001 the factor is 0.998, and the same progress takes about 2,300 steps, more than 100 times as many.
On a fixed epoch budget, a tiny rate also stops before converging, which looks like underfitting; the loss curve shows a smooth but nearly flat decline. Interviewers often ask how to choose the rate: a learning-rate range test, warmup plus a decay schedule, or an adaptive optimizer such as Adam. A rate that is too high fails the opposite way, with a loss that oscillates or diverges.