Model Averaging Weights

Problem

Compute softmax model averaging weights from losses: exp(-(loss_i - min_loss)/T) normalized to sum to 1.

Return uniform weights when temperature <= 0 and [] for empty input.

Hint

Lower temperature concentrates weight on the best-performing model.

case 1

in: [[0.1,0.5],0.2]

expected: [0.8807970779778823,0.11920292202211755]

case 2

in: [[1,1],0.2]

expected: [0.5,0.5]

case 3

in: [[0.5],0]

expected: [1]

case 4

in: [[],1]

expected: []

Starter codePython

import math
def model_averaging_weights(losses, temperature):
    # Your code here
    pass

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