Not supported in CUDA.

Shard a module in-place by updating its parameter dictionary with the sharded parameter dictionary. The ``sharding`` argument can be any call

r"""Applies the sigmoid function. .. math:: \text{Sigmoid}(x) = \sigma(x) = \frac{1}{1 + \exp(-x)}

r"""Applies the Softmax function. Applies :math:`\frac{e^{x_i}}{\sum_j e^{x_j}}` element wise.

r"""Applies the Softshrink activation function. .. math:: \text{softshrink}(x) = \begin{cases} x - \lambda & \text{if } x > \lamb

Any parent in the divider will continue to refer to `x` but any parent not in the divider will refer to a copy of the operation.

The module's state dictionary The module's state dictionary contains any attribute set on the module including parameters in :meth:`M

Applies the hyperbolic tangent function. Simply ``mx.tanh(x)``.