Returns a sorted copy of the flattened array. */

To be used by derived classes to optionally put a parameter on a schedule.

Map dtype to cudnn data type.

Load array from reader in safetensor format */

Recursively return all the non frozen :class:`mlx.core.array` members of this Module as a dict of dicts and lists.

Replace the child modules of this :class:`Module` instance with the provided ones in the dict of dicts and lists. It is the equivalen

Apply a function to all the modules in this instance (including this instance). Args: apply_fn (Callable): The function t

Perform a distributed optimizer step by sharding gradients and optimizer states across ranks. This helper function performs the following steps:

Compute the one-dimensional inverse of `rfft`. */

Load array from reader in .npy format */

Load from device memory into threadgroup memory - without bound checking */

Save array to out stream in .npy format */

r"""Applies the Step Activation Function. This function implements a binary step activation, where the output is set to 1 if the input is gre

* Returns a function which computes the value and gradient of the input * function with respect to a single input array. **/

* Returns indices that partition the flattened array * such that the smaller kth elements are first. **/

Returns indices that sort the flattened array. */