utils.py

from seqeval.metrics import recall_score, precision_score

import numpy as np


def compute_metrics(p, all_labels):
    predictions, labels = p
    predictions = np.argmax(predictions, axis=2)

    # Remove ignored index (special tokens)
    true_predictions = [
        [all_labels[p] for (p, l) in zip(prediction, label) if l != -100]
        for prediction, label in zip(predictions, labels)
    ]
    true_labels = [
        [all_labels[l] for (p, l) in zip(prediction, label) if l != -100]
        for prediction, label in zip(predictions, labels)
    ]

    recall = recall_score(true_labels, true_predictions)
    precision = precision_score(true_labels, true_predictions)
    f1_score = (1 + 5 * 5) * recall * precision / (5 * 5 * precision + recall)

    results = {"recall": recall, "precision": precision, "f1": f1_score}
    return results