EasyNLPnlp-218

VQA Accuracy

Problem

Return the fraction of VQA predictions that match the gold answer after stripping whitespace and lowercasing.

The lists are zipped to the shorter length, and an empty gold list gives 0.0.

Hint

Normalize both sides with strip and lower before comparing.

case 1

in: [["yes","no","blue"],["yes","yes","Blue "]]

expected: 0.6666666666666666

case 2

in: [[],[]]

expected: 0

case 3

in: [["a"],["a","b"]]

expected: 1

case 4

in: [["A "],["a"]]

expected: 1

case 5

in: [["x"],["y"]]

expected: 0

Starter codePython

def vqa_accuracy(predictions, golds):
    # Your code here
    pass

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