d44475866914c19f23c0f8a833951f9989250334,gpytorch/kernels/kernel.py,Kernel,__init__,#Kernel#Any#Any#Any#Any#Any#Any#Any#,88

Before Change

        super(Kernel, self).__init__()
        if active_dims is not None and not torch.is_tensor(active_dims):
            active_dims = torch.tensor(active_dims, dtype=torch.long)
        self.register_buffer("active_dims", active_dims)
        self.ard_num_dims = ard_num_dims
        self.batch_size = batch_size
        self.__has_lengthscale = has_lengthscale
        self._param_transform = param_transform

After Change

    * Default: No lengthscale (i.e. :math:`\Theta` is the identity matrix).

    * Single lengthscale: One lengthscale can be applied to all input dimensions/batches
      (i.e. :math:`\Theta` is a constant diagonal matrix).
      This is controlled by setting `has_lengthscale=True`.

    * ARD: Each input dimension gets its own separate lengthscale
      (i.e. :math:`\Theta` is a non-constant diagonal matrix).
      This is controlled by the `ard_num_dims` keyword argument (as well has `has_lengthscale=True`).

    In batch-mode (i.e. when :math:`x_1` and :math:`x_2` are batches of input matrices), each
    batch of data can have its own lengthscale parameter by setting the `batch_size`
    keyword argument to the appropriate number of batches.

    .. note::

        The :attr:`lengthscale` parameter is parameterized on a log scale to constrain it to be positive.
        You can set a prior on this parameter using the :attr:`lengthscale_prior` argument.

Instances