EasyNLPnlp-093

BLEU Brevity Penalty

Problem

Return the BLEU brevity penalty min(1, exp(1 - ref_len / hyp_len)).

Return 0.0 when hyp_len is 0.

Hint

The penalty only bites when the hypothesis is shorter than the reference.

case 1

in: [10,10]

expected: 1

case 2

in: [5,10]

expected: 0.36787944117144233

case 3

in: [10,5]

expected: 1

case 4

in: [0,5]

expected: 0

case 5

in: [1,1]

expected: 1

Starter codePython

import math
def bleu_brevity_penalty(hyp_len, ref_len):
    # Your code here
    pass

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