Source code for poutyne.framework.metrics.epoch_metrics.base

from abc import ABC, abstractmethod

import torch.nn as nn


class EpochMetric(ABC, nn.Module):
    """
    The abstract class representing a epoch metric which can be accumulated at each batch and calculated at the end
    of the epoch.
    """

    @abstractmethod
    def forward(self, y_pred, y_true) -> None:
        """
        To define the behavior of the metric when called.

        Args:
            y_pred: The prediction of the model.
            y_true: Target to evaluate the model.
        """
        pass

    @abstractmethod
    def get_metric(self):
        """
        Compute and return the metric. Should not modify the state of
        the epoch metric.
        """
        pass

    @abstractmethod
    def reset(self) -> None:
        """
        The information kept for the computation of the metric is cleaned so
        that a new epoch can be done.
        """
        pass