Contrastive Information Gap

Problem

Return the contrastive information gap in bits: the average negative similarity minus the positive similarity, converted from nats to bits by dividing by ln(2).

Hint

A large gap means the positive pair is far more similar than unrelated pairs.

Test cases

case 1

in: [1,[0,0]]

expected: -1.4426950408889634

case 2

in: [0.5,[1,2]]

expected: 1.4426950408889634

case 3

in: [0,[0,0]]

expected: 0

case 4

in: [2,[1]]

expected: -1.4426950408889634

Starter codePython

import math
def contrastive_information_gap(pos_sim, neg_sims):
    # Your code here
    pass

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