tf.experimental.dtensor.name_based_save

Saves name based Tensor into a Checkpoint.

tf.experimental.dtensor.name_based_save(
    mesh: tf.experimental.dtensor.Mesh,
    checkpoint_prefix: Union[str, tf.Tensor],
    name_tensor_dict: Dict[str, Union[ops.Tensor, tf_variables.Variable]]
)

The function prepares the input dictionary to the format of a sharded_save, so that it can take advantage of DTensor SPMD based distributed save.

Same as restore, the function only supports saving on the single mesh.

Args
mesh	The single mesh that all Tensors would be restored to.
checkpoint_prefix	The prefix of checkpoint to be restored.
name_tensor_dict	A ordered dictionary of tensor_names to a DTensor. The DTensor shape/dtype must match the tensors being saved/restored for now.