One of the most common LR schedules. The learning rate decays exponentially at each training step.
lr_t = initial_lr × decay_rate^t- t=0: full learning rate
- Decays smoothly and continuously
- Used in Adam, AdaGrad, etc.
exponential_lr_schedule(initial_lr=0.1, decay_rate=0.9, steps=5) → [0.1, 0.09, 0.081, 0.0729, 0.06561]
Round to **5 decimal places**.
Similar Problems
Test Cases (2 visible · 1 hidden)
Case 1: Basic schedule
Input: exponential_lr_schedule(0.1,0.9,5)
Expected: [0.1, 0.09, 0.081, 0.0729, 0.06561]
Case 2: Halving each step
Input: exponential_lr_schedule(1.0,0.5,3)
Expected: [1.0, 0.5, 0.25]
⌘↵ Run · ⌘⇧↵ Submit