Acceptance Rate Estimate

Problem

Compute the aggregate speculative-decoding acceptance rate from per-round counts: sum(accepted_counts) / sum(proposed_counts).

The signature is acceptance_rate(accepted_counts, proposed_counts). Return a float between 0 and 1.

Hint

Pool all rounds by summing accepted and proposed tokens separately.

case 1

in: [[8,6],[10,10]]

expected: 0.7

case 2

in: [[0],[5]]

expected: 0

case 3

in: [[5],[5]]

expected: 1

case 4

in: [[3,3,3],[4,4,4]]

expected: 0.75

Starter codePython

def acceptance_rate(accepted_counts, proposed_counts):
    # Your code here
    pass

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