Importance Weight Clipping

Problem

Clip importance weights at max_weight, then rescale them so their mean is 1.

Return zeros when the clipped weights sum to 0; empty input returns [].

Hint

Clipping controls variance at the cost of some bias; renormalizing keeps scale stable.

case 1

in: [[1,2,10],3]

expected: [0.5,1,1.5]

case 2

in: [[0,0],5]

expected: [0,0]

case 3

in: [[5],5]

expected: [1]

case 4

in: [[],3]

expected: []

Starter codePython

def clip_and_normalize_weights(weights, max_weight):
    # Your code here
    pass

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