import evaluate _accuracy_metric = evaluate.load("accuracy") def accuracy(predictions, y): unique_values = set(y + predictions) label_mapping = {label: i for i, label in enumerate(unique_values)} true_labels = [label_mapping[value] for value in y] pred_labels = [label_mapping[value] for value in predictions] # Compute accuracy result = _accuracy_metric.compute(predictions=pred_labels, references=true_labels) return result["accuracy"]