c672738b39c8f739551da0d88991a2fedfba3e14,deeppavlov/models/ranking/metrics.py,,rank_response,#Any#Any#,37

Before Change

            if predictions[i][j] in np.arange(labels[i][j]):
                ranks.append(j)
                break
    return np.mean(np.asarray(ranks).astype(float))


@register_metric("loss")

After Change

    num_examples = float(len(y_pred))
    predictions = np.array(y_pred)
    predictions = np.flip(np.argsort(predictions, -1), -1)
    rank_tot = 0
    for el in predictions:
        for i, x in enumerate(el):
            if x == 0:
                rank_tot += i
                break
    return float(rank_tot)/num_examples

@register_metric("r@1_insQA")
def r_at_1_insQA(y_true, y_pred):
    return recall_at_k_insQA(y_true, y_pred, k=1)

Instances