torchrecurrent.NASCell

class torchrecurrent.NASCell(input_size, hidden_size, bias=True, kernel_init=<function xavier_uniform_>, recurrent_kernel_init=<function xavier_uniform_>, bias_init=<function zeros_>, recurrent_bias_init=<function zeros_>, device=None, dtype=None)

A Neural Architecture Search (NAS) cell.

This cell was discovered by architecture search and applies eight parallel gates to input and hidden projections, then composes them through two “layer” computations to update both cell and hidden states:

\[\begin{split}\mathbf{g}(t) &= \mathbf{W}_{ih}\,\mathbf{x}(t) + \mathbf{b}_{ih} + \mathbf{W}_{hh}\,\mathbf{h}(t-1) + \mathbf{b}_{hh}, \\[6pt] [g_0,\dots,g_7] &= \mathrm{chunk}_8\bigl(\mathbf{g}(t)\bigr), \\[6pt] o_k(t) &= \begin{cases} \sigma(g_k), & k\in\{0,2,5,7\},\\ \mathrm{ReLU}(g_k), & k\in\{1,3\},\\ \tanh(g_k), & k\in\{4,6\}, \end{cases} \\[6pt] \ell_1(t) &= \tanh\bigl(o_0\,\circ\,o_1\bigr),\quad \ell_2(t) = \tanh\bigl(o_2 + o_3\bigr), \\[3pt] \ell_3(t) &= \tanh\bigl(o_4\,\circ\,o_5\bigr),\quad \ell_4(t) = \sigma\bigl(o_6 + o_7\bigr), \\[6pt] \tilde{c}(t) &= \tanh\bigl(\ell_1 + c(t-1)\bigr),\quad c(t) = \ell_1\,\circ\,\ell_2, \\[3pt] \ell_5(t) &= \tanh\bigl(\ell_3 + \ell_4\bigr),\quad h(t) = \tanh\bigl(c(t)\,\circ\,\ell_5(t)\bigr).\end{split}\]

Parameters:

input_size (int) – Dimensionality of input vector \(\mathbf{x}(t)\).
hidden_size (int) – Number of hidden units \(\mathbf{h}(t)\).
bias (bool, optional) – If False, disables both input and hidden biases. Default: True.
kernel_init (Callable, optional) – Initializer for input‐to‐hidden weights \(\mathbf{W}_{ih}\). Default: nn.init.xavier_uniform_.
recurrent_kernel_init (Callable, optional) – Initializer for hidden‐to‐hidden weights \(\mathbf{W}_{hh}\). Default: nn.init.xavier_uniform_.
bias_init (Callable, optional) – Initializer for input biases \(\mathbf{b}_{ih}\). Default: nn.init.zeros_.
recurrent_bias_init (Callable, optional) – Initializer for hidden biases \(\mathbf{b}_{hh}\). Default: nn.init.zeros_.
device (torch.device, optional) – Device for all tensors. Default: CPU.
dtype (torch.dtype, optional) – Dtype for all tensors. Default: float32.

Inputs: input, (h, c)

input (Tensor): shape (H_in,) or (N, H_in), where H_in = input_size.
h, c (Tensor): previous hidden & cell; each of shape
(H_out,) or (N, H_out), where H_out = hidden_size. Defaults to zeros if not provided.

Outputs: (h, c)

h (Tensor): next hidden state, same shape as inputs.
c (Tensor): next cell state, same shape as inputs.

Shape:

input: \((N, H_{\mathrm{in}})\) or \((H_{\mathrm{in}})\).
h, c: \((N, H_{\mathrm{out}})\) or \((H_{\mathrm{out}})\).
output: same as h and c.

weight_ih

input‐to‐hidden weights of shape (8*H, I).

Type:: Tensor

weight_hh

hidden‐to‐hidden weights of shape (8*H, H).

Type:: Tensor

bias_ih

input biases of shape (8*H,) if bias=True.

Type:: Tensor

bias_hh

hidden biases of shape (8*H,) if bias=True.

Type:: Tensor

Examples::

>>> cell = NASCell(10, 20)
>>> x = torch.randn(5, 10)
>>> h, c = torch.zeros(20), torch.zeros(20)
>>> for t in range(x.size(0)):
...     h, c = cell(x[t], (h, c))

__init__(input_size, hidden_size, bias=True, kernel_init=<function xavier_uniform_>, recurrent_kernel_init=<function xavier_uniform_>, bias_init=<function zeros_>, recurrent_bias_init=<function zeros_>, device=None, dtype=None)

Initialize internal Module state, shared by both nn.Module and ScriptModule.

Parameters:

input_size (int)
hidden_size (int)
bias (bool)
kernel_init (Callable)
recurrent_kernel_init (Callable)
bias_init (Callable)
recurrent_bias_init (Callable)
device (device | None)
dtype (dtype | None)

Methods

`__init__`(input_size, hidden_size[, bias, ...])	Initialize internal Module state, shared by both nn.Module and ScriptModule.
`add_module`(name, module)	Add a child module to the current module.
`apply`(fn)	Apply `fn` recursively to every submodule (as returned by `.children()`) as well as self.
`bfloat16`()	Casts all floating point parameters and buffers to `bfloat16` datatype.
`buffers`([recurse])	Return an iterator over module buffers.
`children`()	Return an iterator over immediate children modules.
`compile`(args, *kwargs)	Compile this Module's forward using `torch.compile()`.
`cpu`()	Move all model parameters and buffers to the CPU.
`cuda`([device])	Move all model parameters and buffers to the GPU.
`double`()	Casts all floating point parameters and buffers to `double` datatype.
`eval`()	Set the module in evaluation mode.
`extra_repr`()	Return the extra representation of the module.
`float`()	Casts all floating point parameters and buffers to `float` datatype.
`forward`(inp[, state])	Run one step of the recurrent cell.
`get_buffer`(target)	Return the buffer given by `target` if it exists, otherwise throw an error.
`get_extra_state`()	Return any extra state to include in the module's state_dict.
`get_parameter`(target)	Return the parameter given by `target` if it exists, otherwise throw an error.
`get_submodule`(target)	Return the submodule given by `target` if it exists, otherwise throw an error.
`half`()	Casts all floating point parameters and buffers to `half` datatype.
`init_weights`()
`ipu`([device])	Move all model parameters and buffers to the IPU.
`load_state_dict`(state_dict[, strict, assign])	Copy parameters and buffers from `state_dict` into this module and its descendants.
`modules`()	Return an iterator over all modules in the network.
`mtia`([device])	Move all model parameters and buffers to the MTIA.
`named_buffers`([prefix, recurse, ...])	Return an iterator over module buffers, yielding both the name of the buffer as well as the buffer itself.
`named_children`()	Return an iterator over immediate children modules, yielding both the name of the module as well as the module itself.
`named_modules`([memo, prefix, remove_duplicate])	Return an iterator over all modules in the network, yielding both the name of the module as well as the module itself.
`named_parameters`([prefix, recurse, ...])	Return an iterator over module parameters, yielding both the name of the parameter as well as the parameter itself.
`parameters`([recurse])	Return an iterator over module parameters.
`register_backward_hook`(hook)	Register a backward hook on the module.
`register_buffer`(name, tensor[, persistent])	Add a buffer to the module.
`register_forward_hook`(hook, *[, prepend, ...])	Register a forward hook on the module.
`register_forward_pre_hook`(hook, *[, ...])	Register a forward pre-hook on the module.
`register_full_backward_hook`(hook[, prepend])	Register a backward hook on the module.
`register_full_backward_pre_hook`(hook[, prepend])	Register a backward pre-hook on the module.
`register_load_state_dict_post_hook`(hook)	Register a post-hook to be run after module's `load_state_dict()` is called.
`register_load_state_dict_pre_hook`(hook)	Register a pre-hook to be run before module's `load_state_dict()` is called.
`register_module`(name, module)	Alias for `add_module()`.
`register_parameter`(name, param)	Add a parameter to the module.
`register_state_dict_post_hook`(hook)	Register a post-hook for the `state_dict()` method.
`register_state_dict_pre_hook`(hook)	Register a pre-hook for the `state_dict()` method.
`requires_grad_`([requires_grad])	Change if autograd should record operations on parameters in this module.
`set_extra_state`(state)	Set extra state contained in the loaded state_dict.
`set_submodule`(target, module[, strict])	Set the submodule given by `target` if it exists, otherwise throw an error.
`share_memory`()	See `torch.Tensor.share_memory_()`.
`state_dict`(*args[, destination, prefix, ...])	Return a dictionary containing references to the whole state of the module.
`to`(args, *kwargs)	Move and/or cast the parameters and buffers.
`to_empty`(*, device[, recurse])	Move the parameters and buffers to the specified device without copying storage.
`train`([mode])	Set the module in training mode.
`type`(dst_type)	Casts all parameters and buffers to `dst_type`.
`uses_double_state`()	Return True if forward returns (h, c), else just h.
`xpu`([device])	Move all model parameters and buffers to the XPU.
`zero_grad`([set_to_none])	Reset gradients of all model parameters.

Attributes

`T_destination`
`call_super_init`
`dump_patches`
`weight_ih`
`weight_hh`
`bias_ih`
`bias_hh`
`training`