57830b8c78a8f0997f9a7dc61d33a1bf1adb3ca5,pixyz/distributions/flow_distribution.py,InverseTransformedDistribution,sample,#InverseTransformedDistribution#Any#Any#Any#Any#Any#Any#,231

Before Change

        _y = get_dict_values(x_dict, self.cond_var)

        if len(_y) == 0:
            x = self.inverse(_z[0])
        else:
            x = self.inverse(_z[0], y=_y[0])

        output_dict = {self.var[0]: x}

After Change

        p(x=f^{-1}_{flow}(z)),

    where :math:`z \sim p_{prior}(z)`.

    Once initializing, it can be handled as a distribution module.

    Moreover, this distribution can take a conditional variable.

    .. math::

        p(x=f^{-1}_{flow}(z, y)),

    where :math:`z \sim p_{prior}(z)` and :math:`y` is given.

    

Instances