PPO Minibatch Epochs

Problem

Compute the total number of gradient updates in a PPO iteration: ceil(batch_size / minibatch_size) minibatches per epoch times the number of epochs. A non-positive minibatch size returns 0.

Hint

The last minibatch is kept even when the batch does not divide evenly.

Test cases

case 1

in: [64,16,4]

expected: 16

case 2

in: [100,32,3]

expected: 12

case 3

in: [10,10,1]

expected: 1

case 4

in: [10,0,5]

expected: 0

Starter codePython

import math
def ppo_minibatch_updates(batch_size, minibatch_size, epochs):
    # Your code here
    pass

4 test cases · run your code against them in the editor