Adaptive Batch Probability

Problem

Compute per-sample sampling probabilities proportional to gradient norms.

Return [norm_i / sum(norms)]. If the total is zero, return uniform probabilities; an empty list returns [].

Hint

Larger gradients are sampled more often to reduce variance.

case 1

in: [[1,3]]

expected: [0.25,0.75]

case 2

in: [[0,0]]

expected: [0.5,0.5]

case 3

in: [[2]]

expected: [1]

case 4

in: [[]]

expected: []

Starter codePython

def adaptive_batch_probability(grad_norms):
    # Your code here
    pass

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