:py:mod:`n2j.losses.gaussian_nll`
=================================

.. py:module:: n2j.losses.gaussian_nll

.. autoapi-nested-parse::

   Gaussian mixture negative log likelihoods that can be evaluated, for use as
   loss functions, but also generate samples when parameters are given



Module Contents
---------------

Classes
~~~~~~~

.. autoapisummary::

   n2j.losses.gaussian_nll.DiagonalGaussianNLL
   n2j.losses.gaussian_nll.FullRankGaussianNLL
   n2j.losses.gaussian_nll.DoubleGaussianNLL




.. py:class:: DiagonalGaussianNLL(Y_dim)

   Bases: :py:obj:`BaseGaussianNLL`

   The negative log likelihood (NLL) for a single Gaussian with diagonal
   covariance matrix

   `BaseGaussianNLL.__init__` docstring for the parameter description.


   .. py:attribute:: posterior_name
      :value: 'DiagonalGaussianBNNPosterior'

      

   .. py:method:: __call__(pred, target)

      Evaluate the NLL. Must be overridden by subclasses.

      Parameters
      ----------
      pred : torch.Tensor
          raw network output for the predictions
      target : torch.Tensor
          Y labels



   .. py:method:: slice(pred)

      Slice the raw network prediction into meaningful Gaussian parameters

      Parameters
      ----------
      pred : torch.Tensor of shape `[batch_size, self.Y_dim]`
          the network prediction



   .. py:method:: set_trained_pred(pred)


   .. py:method:: sample(mean, std, n_samples)

      Sample from a Gaussian posterior with diagonal covariance matrix
      Parameters
      ----------
      n_samples : int
          how many samples to obtain
      sample_seed : int
          seed for the samples. Default: None
      Returns
      -------
      np.array of shape `[batch_size, n_samples, self.Y_dim]`
          samples



.. py:class:: FullRankGaussianNLL(Y_dim)

   Bases: :py:obj:`BaseGaussianNLL`

   The negative log likelihood (NLL) for a single Gaussian with a full-rank
   covariance matrix

   See `BaseGaussianNLL.__init__` docstring for the parameter description.


   .. py:attribute:: posterior_name
      :value: 'FullRankGaussianBNNPosterior'

      

   .. py:method:: __call__(pred, target)

      Evaluate the NLL. Must be overridden by subclasses.

      Parameters
      ----------
      pred : torch.Tensor
          raw network output for the predictions
      target : torch.Tensor
          Y labels



   .. py:method:: slice(pred)

      Slice the raw network prediction into meaningful Gaussian parameters

      Parameters
      ----------
      pred : torch.Tensor of shape `[batch_size, self.Y_dim]`
          the network prediction



   .. py:method:: set_trained_pred(pred)


   .. py:method:: sample(mean, std, n_samples)



.. py:class:: DoubleGaussianNLL(Y_dim)

   Bases: :py:obj:`BaseGaussianNLL`

   The negative log likelihood (NLL) for a mixture of two Gaussians, each
   with a full but constrained as low-rank plus diagonal covariance

   Only rank 2 is currently supported. `BaseGaussianNLL.__init__` docstring
   for the parameter description.


   .. py:attribute:: posterior_name
      :value: 'DoubleGaussianBNNPosterior'

      

   .. py:method:: __call__(pred, target)

      Evaluate the NLL. Must be overridden by subclasses.

      Parameters
      ----------
      pred : torch.Tensor
          raw network output for the predictions
      target : torch.Tensor
          Y labels



   .. py:method:: slice(pred)

      Slice the raw network prediction into meaningful Gaussian parameters

      Parameters
      ----------
      pred : torch.Tensor of shape `[batch_size, self.Y_dim]`
          the network prediction



   .. py:method:: set_trained_pred(pred)


   .. py:method:: sample(mean, std, n_samples)

      Sample from a mixture of two Gaussians, each with a full covariance

      Parameters
      ----------
      n_samples : int
          how many samples to obtain
      sample_seed : int
          seed for the samples. Default: None
      Returns
      -------
      np.array of shape `[self.batch_size, n_samples, self.Y_dim]`
          samples