utils.py

from collections import Counter
import numpy as np

def normalize_answer(s):
    """Lower text and remove punctuation, articles and extra whitespace."""
    import re
    import string

    def remove_articles(text):
        return re.sub(r'\b(a|an|the)\b', ' ', text)

    def white_space_fix(text):
        return ' '.join(text.split())

    def remove_punc(text):
        exclude = set(string.punctuation)
        return ''.join(ch for ch in text if ch not in exclude)

    def lower(text):
        return text.lower()

    return white_space_fix(remove_articles(remove_punc(lower(s))))


def compute_f1(predictions, references):
    f1_scores = []
    for prediction, reference in zip(predictions, references):
        prediction_tokens = normalize_answer(prediction).split()
        ground_truth_tokens = normalize_answer(reference).split()
        common = Counter(prediction_tokens) & Counter(ground_truth_tokens)
        num_same = sum(common.values())

        if num_same == 0:
            f1_scores.append(0)
        else:
            precision = 1.0 * num_same / len(prediction_tokens)
            recall = 1.0 * num_same / len(ground_truth_tokens)
            f1_scores.append((2 * precision * recall) / (precision + recall))

    return np.mean(f1_scores)