RND Normalized Bonus

Problem

Normalize RND prediction errors into a bonus by z-scoring them.

Return (e - mean) / std using population statistics; if the standard deviation is zero return zeros, and an empty batch returns [].

Hint

Normalizing novelty across a batch keeps the intrinsic reward scale stable.

case 1

in: [[1,2,3]]

expected: [-1.224744871391589,0,1.224744871391589]

case 2

in: [[5,5]]

expected: [0,0]

case 3

in: [[]]

expected: []

case 4

in: [[-1,0,1]]

expected: [-1.224744871391589,0,1.224744871391589]

Starter codePython

import math
def rnd_normalized_bonus(errors):
    # Your code here
    pass

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