Ensemble Critic Variance

Problem

Compute the per-action variance of Q values across an ensemble.

predictions[m][a] is action a's value from ensemble member m; return the population variance across members for each action. An empty ensemble returns [].

Hint

Disagreement across independent critics is a cheap uncertainty signal.

Test cases

case 1

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

expected: [1,0]

case 2

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

expected: [0.6666666666666666]

case 3

in: [[]]

expected: []

Starter codePython

def ensemble_critic_variance(predictions):
    # Your code here
    pass

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