API Reference

Nodes

exception hierarchicalsoftmax.nodes.AlreadyIndexedError

Raised when set_indexes run more than once on a node.

with_traceback(): Exception.with_traceback(tb) – set self.__traceback__ to tb and return self.

exception hierarchicalsoftmax.nodes.IndexNotSetError

Raised when set_indexes not set for the SoftmaxNode root.

with_traceback(): Exception.with_traceback(tb) – set self.__traceback__ to tb and return self.

exception hierarchicalsoftmax.nodes.ReadOnlyError

Raised when trying to edit a SoftmaxNode tree after it has been set to read only.

with_traceback(): Exception.with_traceback(tb) – set self.__traceback__ to tb and return self.

class hierarchicalsoftmax.nodes.SoftmaxNode(*args, alpha: float = 1.0, weight=None, label_smoothing: float = 0.0, gamma: Optional[float] = None, readonly: bool = False, **kwargs)

Creates a hierarchical tree to perform a softmax at each level.

property ancestors

All parent nodes and their parent nodes.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.ancestors
()
>>> marc.ancestors
(Node('/Udo'),)
>>> lian.ancestors
(Node('/Udo'), Node('/Udo/Marc'))

property anchestors

All parent nodes and their parent nodes - see ancestors.

The attribute anchestors is just a typo of ancestors. Please use ancestors. This attribute will be removed in the 3.0.0 release.

property children

All child nodes.

>>> from anytree import Node
>>> n = Node("n")
>>> a = Node("a", parent=n)
>>> b = Node("b", parent=n)
>>> c = Node("c", parent=n)
>>> n.children
(Node('/n/a'), Node('/n/b'), Node('/n/c'))

Modifying the children attribute modifies the tree.

Detach

The children attribute can be updated by setting to an iterable.

>>> n.children = [a, b]
>>> n.children
(Node('/n/a'), Node('/n/b'))

Node c is removed from the tree. In case of an existing reference, the node c does not vanish and is the root of its own tree.

>>> c
Node('/c')

Attach

>>> d = Node("d")
>>> d
Node('/d')
>>> n.children = [a, b, d]
>>> n.children
(Node('/n/a'), Node('/n/b'), Node('/n/d'))
>>> d
Node('/n/d')

Duplicate

A node can just be the children once. Duplicates cause a TreeError:

>>> n.children = [a, b, d, a]
Traceback (most recent call last):
    ...
anytree.node.exceptions.TreeError: Cannot add node Node('/n/a') multiple times as child.

property depth

Number of edges to the root Node.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.depth
0
>>> marc.depth
1
>>> lian.depth
2

property descendants

All child nodes and all their child nodes.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> loui = Node("Loui", parent=marc)
>>> soe = Node("Soe", parent=lian)
>>> udo.descendants
(Node('/Udo/Marc'), Node('/Udo/Marc/Lian'), Node('/Udo/Marc/Lian/Soe'), Node('/Udo/Marc/Loui'))
>>> marc.descendants
(Node('/Udo/Marc/Lian'), Node('/Udo/Marc/Lian/Soe'), Node('/Udo/Marc/Loui'))
>>> lian.descendants
(Node('/Udo/Marc/Lian/Soe'),)

get_child_by_name(name: str) → hierarchicalsoftmax.nodes.SoftmaxNode

Returns the child node that has the same name as what is given.

Parameters: name (str) – The name of the child node requested.
Returns: The child node that has the same name as what is given. If not child node exists with this name then None is returned.
Return type: SoftmaxNode

get_node_ids(nodes: List) → List[int]

Gets the index values for descendant nodes.

This should only be used for root nodes. If set_indexes has been yet called on this object then it is performed as part of this function call.

Parameters: nodes (List) – A list of descendant nodes.
Returns: A list of indexes for the descendant nodes requested.
Return type: List[int]

get_node_ids_tensor(nodes: List) → torch.Tensor

Gets the index values for descendant nodes.

This should only be used for root nodes. If set_indexes has been yet called on this object then it is performed as part of this function call.

Parameters: nodes (List) – A list of descendant nodes.
Returns: A tensor which contains the indexes for the descendant nodes requested.
Return type: torch.Tensor

property height

Number of edges on the longest path to a leaf Node.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.height
2
>>> marc.height
1
>>> lian.height
0

property is_leaf

Node has no children (External Node).

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.is_leaf
False
>>> marc.is_leaf
False
>>> lian.is_leaf
True

property is_root

Node is tree root.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.is_root
True
>>> marc.is_root
False
>>> lian.is_root
False

iter_path_reverse()

Iterate up the tree from the current node.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> for node in udo.iter_path_reverse():
...     print(node)
Node('/Udo')
>>> for node in marc.iter_path_reverse():
...     print(node)
Node('/Udo/Marc')
Node('/Udo')
>>> for node in lian.iter_path_reverse():
...     print(node)
Node('/Udo/Marc/Lian')
Node('/Udo/Marc')
Node('/Udo')

property leaves

Tuple of all leaf nodes.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> loui = Node("Loui", parent=marc)
>>> lazy = Node("Lazy", parent=marc)
>>> udo.leaves
(Node('/Udo/Marc/Lian'), Node('/Udo/Marc/Loui'), Node('/Udo/Marc/Lazy'))
>>> marc.leaves
(Node('/Udo/Marc/Lian'), Node('/Udo/Marc/Loui'), Node('/Udo/Marc/Lazy'))

level_order_group_iter(depth=None, **kwargs) → anytree.iterators.levelordergroupiter.LevelOrderGroupIter

Returns a level-order iterator with grouping starting at this node.

https://anytree.readthedocs.io/en/latest/api/anytree.iterators.html#anytree.iterators.levelordergroupiter.LevelOrderGroupIter

level_order_iter(depth=None, **kwargs) → anytree.iterators.levelorderiter.LevelOrderIter

Returns a level-order iterator.

See https://anytree.readthedocs.io/en/latest/api/anytree.iterators.html#anytree.iterators.levelorderiter.LevelOrderIter

property parent

Parent Node.

On set, the node is detached from any previous parent node and attached to the new node.

>>> from anytree import Node, RenderTree
>>> udo = Node("Udo")
>>> marc = Node("Marc")
>>> lian = Node("Lian", parent=marc)
>>> print(RenderTree(udo))
Node('/Udo')
>>> print(RenderTree(marc))
Node('/Marc')
└── Node('/Marc/Lian')

Attach

>>> marc.parent = udo
>>> print(RenderTree(udo))
Node('/Udo')
└── Node('/Udo/Marc')
    └── Node('/Udo/Marc/Lian')

Detach

To make a node to a root node, just set this attribute to None.

>>> marc.is_root
False
>>> marc.parent = None
>>> marc.is_root
True

property path

Path of this Node.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.path
(Node('/Udo'),)
>>> marc.path
(Node('/Udo'), Node('/Udo/Marc'))
>>> lian.path
(Node('/Udo'), Node('/Udo/Marc'), Node('/Udo/Marc/Lian'))

post_order_iter(depth=None, **kwargs) → anytree.iterators.postorderiter.PostOrderIter

Returns a post-order iterator.

See https://anytree.readthedocs.io/en/latest/api/anytree.iterators.html#anytree.iterators.postorderiter.PostOrderIter

pre_order_iter(depth=None, **kwargs) → anytree.iterators.preorderiter.PreOrderIter

Returns a pre-order iterator.

See https://anytree.readthedocs.io/en/latest/api/anytree.iterators.html#anytree.iterators.preorderiter.PreOrderIter

render(attr: Optional[str] = None, print: bool = False, filepath: Optional[Union[str, pathlib.Path]] = None, **kwargs) → anytree.render.RenderTree

Renders this node and all its descendants in a tree format.

Parameters

attr (str, optional) – An attribute to print for this rendering of the tree. If None, then the name of each node is used.
print (bool) – Whether or not the tree should be printed. Defaults to False.
filepath – (str, Path, optional): A path to save the tree to using graphviz. Requires graphviz to be installed.

Returns

The tree rendered by anytree.

Return type

RenderTree

render_equal(string_representation: str, **kwargs) → bool

Checks if the string representation of this node and its descendants matches the given string.

Parameters: string_representation (str) – The string representation to compare to.

property root

Tree Root Node.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> udo.root
Node('/Udo')
>>> marc.root
Node('/Udo')
>>> lian.root
Node('/Udo')

set_indexes(index_in_parent: Optional[int] = None, current_index: int = 0) → int

Sets all the indexes for this node and its descendants so that each node can be referenced by the root.

This should be called without arguments only on the root of a hierarchy tree. After calling this function the tree from the root down will be read only.

Parameters

index_in_parent (int, optional) – The index of this node in the parent’s list of children. Defaults to None which is appropriate for the root of a tree.
current_index (int, optional) – An index value for the root node to reference this node. Defaults to 0 which is appropriate for the root of a tree.

Returns

Returns the current_index

Return type

int

set_indexes_if_unset() → None

Calls set_indexes if it has not been called yet.

This is only appropriate for the root node.

property siblings

Tuple of nodes with the same parent.

>>> from anytree import Node
>>> udo = Node("Udo")
>>> marc = Node("Marc", parent=udo)
>>> lian = Node("Lian", parent=marc)
>>> loui = Node("Loui", parent=marc)
>>> lazy = Node("Lazy", parent=marc)
>>> udo.siblings
()
>>> marc.siblings
()
>>> lian.siblings
(Node('/Udo/Marc/Loui'), Node('/Udo/Marc/Lazy'))
>>> loui.siblings
(Node('/Udo/Marc/Lian'), Node('/Udo/Marc/Lazy'))

zig_zag_group_iter(depth=None, **kwargs) → anytree.iterators.zigzaggroupiter.ZigZagGroupIter

Returns a zig-zag iterator with grouping starting at this node.

https://anytree.readthedocs.io/en/latest/api/anytree.iterators.html#anytree.iterators.zigzaggroupiter.ZigZagGroupIter

Layers

exception hierarchicalsoftmax.layers.HierarchicalSoftmaxLayerError

with_traceback(): Exception.with_traceback(tb) – set self.__traceback__ to tb and return self.

class hierarchicalsoftmax.layers.HierarchicalSoftmaxLazyLinear(root: hierarchicalsoftmax.nodes.SoftmaxNode, out_features=None, **kwargs)

Creates a lazy linear layer designed to be the final layer in a neural network model that produces unnormalized scores given to HierarchicalSoftmaxLoss.

The out_features value is set internally from root.layer_size and cannot be given as an argument. The in_features will be inferred from the previous layer at runtime.

add_module(name: str, module: Optional[torch.nn.modules.module.Module]) → None

Adds a child module to the current module.

The module can be accessed as an attribute using the given name.

Parameters

name (str) – name of the child module. The child module can be accessed from this module using the given name
module (Module) – child module to be added to the module.

apply(fn: Callable[[torch.nn.modules.module.Module], None]) → torch.nn.modules.module.T

Applies fn recursively to every submodule (as returned by .children()) as well as self. Typical use includes initializing the parameters of a model (see also nn-init-doc).

Parameters: fn (Module -> None) – function to be applied to each submodule
Returns: self
Return type: Module

Example:

>>> @torch.no_grad()
>>> def init_weights(m):
>>>     print(m)
>>>     if type(m) == nn.Linear:
>>>         m.weight.fill_(1.0)
>>>         print(m.weight)
>>> net = nn.Sequential(nn.Linear(2, 2), nn.Linear(2, 2))
>>> net.apply(init_weights)
Linear(in_features=2, out_features=2, bias=True)
Parameter containing:
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
Linear(in_features=2, out_features=2, bias=True)
Parameter containing:
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
)

bfloat16() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to bfloat16 datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

buffers(recurse: bool = True) → Iterator[torch.Tensor]

Returns an iterator over module buffers.

Parameters: recurse (bool) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module.
Yields: torch.Tensor – module buffer

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for buf in model.buffers():
>>>     print(type(buf), buf.size())
<class 'torch.Tensor'> (20L,)
<class 'torch.Tensor'> (20L, 1L, 5L, 5L)

children() → Iterator[torch.nn.modules.module.Module]

Returns an iterator over immediate children modules.

Yields: Module – a child module

cls_to_become: alias of torch.nn.modules.linear.Linear

cpu() → torch.nn.modules.module.T

Moves all model parameters and buffers to the CPU.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

cuda(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the GPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on GPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

double() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to double datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

eval() → torch.nn.modules.module.T

Sets the module in evaluation mode.

This has any effect only on certain modules. See documentations of particular modules for details of their behaviors in training/evaluation mode, if they are affected, e.g. Dropout, BatchNorm, etc.

This is equivalent with self.train(False).

See locally-disable-grad-doc for a comparison between .eval() and several similar mechanisms that may be confused with it.

Returns: self
Return type: Module

extra_repr() → str

Set the extra representation of the module

To print customized extra information, you should re-implement this method in your own modules. Both single-line and multi-line strings are acceptable.

float() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to float datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

forward(x) → hierarchicalsoftmax.tensors.LazyLinearTensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

get_buffer(target: str) → torch.Tensor

Returns the buffer given by target if it exists, otherwise throws an error.

See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target.

Parameters: target – The fully-qualified string name of the buffer to look for. (See get_submodule for how to specify a fully-qualified string.)
Returns: The buffer referenced by target
Return type: torch.Tensor
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not a buffer

get_extra_state() → Any

Returns any extra state to include in the module’s state_dict. Implement this and a corresponding set_extra_state() for your module if you need to store extra state. This function is called when building the module’s state_dict().

Note that extra state should be pickleable to ensure working serialization of the state_dict. We only provide provide backwards compatibility guarantees for serializing Tensors; other objects may break backwards compatibility if their serialized pickled form changes.

Returns: Any extra state to store in the module’s state_dict
Return type: object

get_parameter(target: str) → torch.nn.parameter.Parameter

Returns the parameter given by target if it exists, otherwise throws an error.

See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target.

Parameters: target – The fully-qualified string name of the Parameter to look for. (See get_submodule for how to specify a fully-qualified string.)
Returns: The Parameter referenced by target
Return type: torch.nn.Parameter
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not an nn.Parameter

get_submodule(target: str) → torch.nn.modules.module.Module

Returns the submodule given by target if it exists, otherwise throws an error.

For example, let’s say you have an nn.Module A that looks like this:

A(
    (net_b): Module(
        (net_c): Module(
            (conv): Conv2d(16, 33, kernel_size=(3, 3), stride=(2, 2))
        )
        (linear): Linear(in_features=100, out_features=200, bias=True)
    )
)

(The diagram shows an nn.Module A. A has a nested submodule net_b, which itself has two submodules net_c and linear. net_c then has a submodule conv.)

To check whether or not we have the linear submodule, we would call get_submodule("net_b.linear"). To check whether we have the conv submodule, we would call get_submodule("net_b.net_c.conv").

The runtime of get_submodule is bounded by the degree of module nesting in target. A query against named_modules achieves the same result, but it is O(N) in the number of transitive modules. So, for a simple check to see if some submodule exists, get_submodule should always be used.

Parameters: target – The fully-qualified string name of the submodule to look for. (See above example for how to specify a fully-qualified string.)
Returns: The submodule referenced by target
Return type: torch.nn.Module
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not an nn.Module

half() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to half datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

has_uninitialized_params(): Check if a module has parameters that are not initialized

initialize_parameters(input) → None: Initialize parameters according to the input batch properties. This adds an interface to isolate parameter initialization from the forward pass when doing parameter shape inference.

ipu(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the IPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on IPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

load_state_dict(state_dict: Mapping[str, Any], strict: bool = True)

Copies parameters and buffers from state_dict into this module and its descendants. If strict is True, then the keys of state_dict must exactly match the keys returned by this module’s state_dict() function.

Parameters

state_dict (dict) – a dict containing parameters and persistent buffers.
strict (bool, optional) – whether to strictly enforce that the keys in state_dict match the keys returned by this module’s state_dict() function. Default: True

Returns

missing_keys is a list of str containing the missing keys
unexpected_keys is a list of str containing the unexpected keys

Return type

NamedTuple with missing_keys and unexpected_keys fields

Note

If a parameter or buffer is registered as None and its corresponding key exists in state_dict, load_state_dict() will raise a RuntimeError.

modules() → Iterator[torch.nn.modules.module.Module]

Returns an iterator over all modules in the network.

Yields: Module – a module in the network

Note

Duplicate modules are returned only once. In the following example, l will be returned only once.

Example:

>>> l = nn.Linear(2, 2)
>>> net = nn.Sequential(l, l)
>>> for idx, m in enumerate(net.modules()):
...     print(idx, '->', m)

0 -> Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
)
1 -> Linear(in_features=2, out_features=2, bias=True)

named_buffers(prefix: str = '', recurse: bool = True) → Iterator[Tuple[str, torch.Tensor]]

Returns an iterator over module buffers, yielding both the name of the buffer as well as the buffer itself.

Parameters

prefix (str) – prefix to prepend to all buffer names.
recurse (bool) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module.

Yields

(str, torch.Tensor) – Tuple containing the name and buffer

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, buf in self.named_buffers():
>>>    if name in ['running_var']:
>>>        print(buf.size())

named_children() → Iterator[Tuple[str, torch.nn.modules.module.Module]]

Returns an iterator over immediate children modules, yielding both the name of the module as well as the module itself.

Yields: (str, Module) – Tuple containing a name and child module

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, module in model.named_children():
>>>     if name in ['conv4', 'conv5']:
>>>         print(module)

named_modules(memo: Optional[Set[torch.nn.modules.module.Module]] = None, prefix: str = '', remove_duplicate: bool = True)

Returns an iterator over all modules in the network, yielding both the name of the module as well as the module itself.

Parameters

memo – a memo to store the set of modules already added to the result
prefix – a prefix that will be added to the name of the module
remove_duplicate – whether to remove the duplicated module instances in the result or not

Yields

(str, Module) – Tuple of name and module

Note

Duplicate modules are returned only once. In the following example, l will be returned only once.

Example:

>>> l = nn.Linear(2, 2)
>>> net = nn.Sequential(l, l)
>>> for idx, m in enumerate(net.named_modules()):
...     print(idx, '->', m)

0 -> ('', Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
))
1 -> ('0', Linear(in_features=2, out_features=2, bias=True))

named_parameters(prefix: str = '', recurse: bool = True) → Iterator[Tuple[str, torch.nn.parameter.Parameter]]

Returns an iterator over module parameters, yielding both the name of the parameter as well as the parameter itself.

Parameters

prefix (str) – prefix to prepend to all parameter names.
recurse (bool) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module.

Yields

(str, Parameter) – Tuple containing the name and parameter

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, param in self.named_parameters():
>>>    if name in ['bias']:
>>>        print(param.size())

parameters(recurse: bool = True) → Iterator[torch.nn.parameter.Parameter]

Returns an iterator over module parameters.

This is typically passed to an optimizer.

Parameters: recurse (bool) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module.
Yields: Parameter – module parameter

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for param in model.parameters():
>>>     print(type(param), param.size())
<class 'torch.Tensor'> (20L,)
<class 'torch.Tensor'> (20L, 1L, 5L, 5L)

register_backward_hook(hook: Callable[[torch.nn.modules.module.Module, Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[None, torch.Tensor]]) → torch.utils.hooks.RemovableHandle

Registers a backward hook on the module.

This function is deprecated in favor of register_full_backward_hook() and the behavior of this function will change in future versions.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_buffer(name: str, tensor: Optional[torch.Tensor], persistent: bool = True) → None

Adds a buffer to the module.

This is typically used to register a buffer that should not to be considered a model parameter. For example, BatchNorm’s running_mean is not a parameter, but is part of the module’s state. Buffers, by default, are persistent and will be saved alongside parameters. This behavior can be changed by setting persistent to False. The only difference between a persistent buffer and a non-persistent buffer is that the latter will not be a part of this module’s state_dict.

Buffers can be accessed as attributes using given names.

Parameters

name (str) – name of the buffer. The buffer can be accessed from this module using the given name
tensor (Tensor or None) – buffer to be registered. If None, then operations that run on buffers, such as cuda, are ignored. If None, the buffer is not included in the module’s state_dict.
persistent (bool) – whether the buffer is part of this module’s state_dict.

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> self.register_buffer('running_mean', torch.zeros(num_features))

register_forward_hook(hook: Callable[[...], None]) → torch.utils.hooks.RemovableHandle

Registers a forward hook on the module.

The hook will be called every time after forward() has computed an output. It should have the following signature:

hook(module, input, output) -> None or modified output

The input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the output. It can modify the input inplace but it will not have effect on forward since this is called after forward() is called.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_forward_pre_hook(hook: Callable[[...], None]) → torch.utils.hooks.RemovableHandle

Registers a forward pre-hook on the module.

The hook will be called every time before forward() is invoked. It should have the following signature:

hook(module, input) -> None or modified input

The input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the input. User can either return a tuple or a single modified value in the hook. We will wrap the value into a tuple if a single value is returned(unless that value is already a tuple).

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_full_backward_hook(hook: Callable[[torch.nn.modules.module.Module, Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[None, torch.Tensor]]) → torch.utils.hooks.RemovableHandle

Registers a backward hook on the module.

The hook will be called every time the gradients with respect to a module are computed, i.e. the hook will execute if and only if the gradients with respect to module outputs are computed. The hook should have the following signature:

hook(module, grad_input, grad_output) -> tuple(Tensor) or None

The grad_input and grad_output are tuples that contain the gradients with respect to the inputs and outputs respectively. The hook should not modify its arguments, but it can optionally return a new gradient with respect to the input that will be used in place of grad_input in subsequent computations. grad_input will only correspond to the inputs given as positional arguments and all kwarg arguments are ignored. Entries in grad_input and grad_output will be None for all non-Tensor arguments.

For technical reasons, when this hook is applied to a Module, its forward function will receive a view of each Tensor passed to the Module. Similarly the caller will receive a view of each Tensor returned by the Module’s forward function.

Warning

Modifying inputs or outputs inplace is not allowed when using backward hooks and will raise an error.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_load_state_dict_post_hook(hook)

Registers a post hook to be run after module’s load_state_dict is called.

It should have the following signature::: hook(module, incompatible_keys) -> None

The module argument is the current module that this hook is registered on, and the incompatible_keys argument is a NamedTuple consisting of attributes missing_keys and unexpected_keys. missing_keys is a list of str containing the missing keys and unexpected_keys is a list of str containing the unexpected keys.

The given incompatible_keys can be modified inplace if needed.

Note that the checks performed when calling load_state_dict() with strict=True are affected by modifications the hook makes to missing_keys or unexpected_keys, as expected. Additions to either set of keys will result in an error being thrown when strict=True, and clearning out both missing and unexpected keys will avoid an error.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_module(name: str, module: Optional[torch.nn.modules.module.Module]) → None: Alias for add_module().

register_parameter(name: str, param: Optional[torch.nn.parameter.Parameter]) → None

Adds a parameter to the module.

The parameter can be accessed as an attribute using given name.

Parameters

name (str) – name of the parameter. The parameter can be accessed from this module using the given name
param (Parameter or None) – parameter to be added to the module. If None, then operations that run on parameters, such as cuda, are ignored. If None, the parameter is not included in the module’s state_dict.

requires_grad_(requires_grad: bool = True) → torch.nn.modules.module.T

Change if autograd should record operations on parameters in this module.

This method sets the parameters’ requires_grad attributes in-place.

This method is helpful for freezing part of the module for finetuning or training parts of a model individually (e.g., GAN training).

See locally-disable-grad-doc for a comparison between .requires_grad_() and several similar mechanisms that may be confused with it.

Parameters: requires_grad (bool) – whether autograd should record operations on parameters in this module. Default: True.
Returns: self
Return type: Module

set_extra_state(state: Any)

This function is called from load_state_dict() to handle any extra state found within the state_dict. Implement this function and a corresponding get_extra_state() for your module if you need to store extra state within its state_dict.

Parameters: state (dict) – Extra state from the state_dict

share_memory() → torch.nn.modules.module.T: See torch.Tensor.share_memory_()

state_dict(*args, destination=None, prefix='', keep_vars=False)

Returns a dictionary containing references to the whole state of the module.

Both parameters and persistent buffers (e.g. running averages) are included. Keys are corresponding parameter and buffer names. Parameters and buffers set to None are not included.

Note

The returned object is a shallow copy. It contains references to the module’s parameters and buffers.

Warning

Currently state_dict() also accepts positional arguments for destination, prefix and keep_vars in order. However, this is being deprecated and keyword arguments will be enforced in future releases.

Warning

Please avoid the use of argument destination as it is not designed for end-users.

Parameters

destination (dict, optional) – If provided, the state of module will be updated into the dict and the same object is returned. Otherwise, an OrderedDict will be created and returned. Default: None.
prefix (str, optional) – a prefix added to parameter and buffer names to compose the keys in state_dict. Default: ''.
keep_vars (bool, optional) – by default the Tensor s returned in the state dict are detached from autograd. If it’s set to True, detaching will not be performed. Default: False.

Returns

a dictionary containing a whole state of the module

Return type

dict

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> module.state_dict().keys()
['bias', 'weight']

to(*args, **kwargs)

Moves and/or casts the parameters and buffers.

This can be called as

to(device=None, dtype=None, non_blocking=False)

to(dtype, non_blocking=False)

to(tensor, non_blocking=False)

to(memory_format=torch.channels_last)

Its signature is similar to torch.Tensor.to(), but only accepts floating point or complex dtypes. In addition, this method will only cast the floating point or complex parameters and buffers to dtype (if given). The integral parameters and buffers will be moved device, if that is given, but with dtypes unchanged. When non_blocking is set, it tries to convert/move asynchronously with respect to the host if possible, e.g., moving CPU Tensors with pinned memory to CUDA devices.

See below for examples.

Note

This method modifies the module in-place.

Parameters

device (torch.device) – the desired device of the parameters and buffers in this module
dtype (torch.dtype) – the desired floating point or complex dtype of the parameters and buffers in this module
tensor (torch.Tensor) – Tensor whose dtype and device are the desired dtype and device for all parameters and buffers in this module
memory_format (torch.memory_format) – the desired memory format for 4D parameters and buffers in this module (keyword only argument)

Returns

self

Return type

Module

Examples:

>>> # xdoctest: +IGNORE_WANT("non-deterministic")
>>> linear = nn.Linear(2, 2)
>>> linear.weight
Parameter containing:
tensor([[ 0.1913, -0.3420],
        [-0.5113, -0.2325]])
>>> linear.to(torch.double)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1913, -0.3420],
        [-0.5113, -0.2325]], dtype=torch.float64)
>>> # xdoctest: +REQUIRES(env:TORCH_DOCTEST_CUDA1)
>>> gpu1 = torch.device("cuda:1")
>>> linear.to(gpu1, dtype=torch.half, non_blocking=True)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1914, -0.3420],
        [-0.5112, -0.2324]], dtype=torch.float16, device='cuda:1')
>>> cpu = torch.device("cpu")
>>> linear.to(cpu)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1914, -0.3420],
        [-0.5112, -0.2324]], dtype=torch.float16)

>>> linear = nn.Linear(2, 2, bias=None).to(torch.cdouble)
>>> linear.weight
Parameter containing:
tensor([[ 0.3741+0.j,  0.2382+0.j],
        [ 0.5593+0.j, -0.4443+0.j]], dtype=torch.complex128)
>>> linear(torch.ones(3, 2, dtype=torch.cdouble))
tensor([[0.6122+0.j, 0.1150+0.j],
        [0.6122+0.j, 0.1150+0.j],
        [0.6122+0.j, 0.1150+0.j]], dtype=torch.complex128)

to_empty(*, device: Union[str, torch.device]) → torch.nn.modules.module.T

Moves the parameters and buffers to the specified device without copying storage.

Parameters: device (torch.device) – The desired device of the parameters and buffers in this module.
Returns: self
Return type: Module

train(mode: bool = True) → torch.nn.modules.module.T

Sets the module in training mode.

This has any effect only on certain modules. See documentations of particular modules for details of their behaviors in training/evaluation mode, if they are affected, e.g. Dropout, BatchNorm, etc.

Parameters: mode (bool) – whether to set training mode (True) or evaluation mode (False). Default: True.
Returns: self
Return type: Module

type(dst_type: Union[torch.dtype, str]) → torch.nn.modules.module.T

Casts all parameters and buffers to dst_type.

Note

This method modifies the module in-place.

Parameters: dst_type (type or string) – the desired type
Returns: self
Return type: Module

xpu(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the XPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on XPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

zero_grad(set_to_none: bool = False) → None

Sets gradients of all model parameters to zero. See similar function under torch.optim.Optimizer for more context.

Parameters: set_to_none (bool) – instead of setting to zero, set the grads to None. See torch.optim.Optimizer.zero_grad() for details.

class hierarchicalsoftmax.layers.HierarchicalSoftmaxLinear(root: hierarchicalsoftmax.nodes.SoftmaxNode, out_features=None, **kwargs)

Creates a linear layer designed to be the final layer in a neural network model that produces unnormalized scores given to HierarchicalSoftmaxLoss.

The out_features value is set internally from root.layer_size and cannot be given as an argument.

add_module(name: str, module: Optional[torch.nn.modules.module.Module]) → None

Adds a child module to the current module.

The module can be accessed as an attribute using the given name.

Parameters

name (str) – name of the child module. The child module can be accessed from this module using the given name
module (Module) – child module to be added to the module.

apply(fn: Callable[[torch.nn.modules.module.Module], None]) → torch.nn.modules.module.T

Applies fn recursively to every submodule (as returned by .children()) as well as self. Typical use includes initializing the parameters of a model (see also nn-init-doc).

Parameters: fn (Module -> None) – function to be applied to each submodule
Returns: self
Return type: Module

Example:

>>> @torch.no_grad()
>>> def init_weights(m):
>>>     print(m)
>>>     if type(m) == nn.Linear:
>>>         m.weight.fill_(1.0)
>>>         print(m.weight)
>>> net = nn.Sequential(nn.Linear(2, 2), nn.Linear(2, 2))
>>> net.apply(init_weights)
Linear(in_features=2, out_features=2, bias=True)
Parameter containing:
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
Linear(in_features=2, out_features=2, bias=True)
Parameter containing:
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
)

bfloat16() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to bfloat16 datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

buffers(recurse: bool = True) → Iterator[torch.Tensor]

Returns an iterator over module buffers.

Parameters: recurse (bool) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module.
Yields: torch.Tensor – module buffer

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for buf in model.buffers():
>>>     print(type(buf), buf.size())
<class 'torch.Tensor'> (20L,)
<class 'torch.Tensor'> (20L, 1L, 5L, 5L)

children() → Iterator[torch.nn.modules.module.Module]

Returns an iterator over immediate children modules.

Yields: Module – a child module

cpu() → torch.nn.modules.module.T

Moves all model parameters and buffers to the CPU.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

cuda(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the GPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on GPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

double() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to double datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

eval() → torch.nn.modules.module.T

Sets the module in evaluation mode.

This has any effect only on certain modules. See documentations of particular modules for details of their behaviors in training/evaluation mode, if they are affected, e.g. Dropout, BatchNorm, etc.

This is equivalent with self.train(False).

See locally-disable-grad-doc for a comparison between .eval() and several similar mechanisms that may be confused with it.

Returns: self
Return type: Module

extra_repr() → str

Set the extra representation of the module

To print customized extra information, you should re-implement this method in your own modules. Both single-line and multi-line strings are acceptable.

float() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to float datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

forward(x) → hierarchicalsoftmax.tensors.LazyLinearTensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

get_buffer(target: str) → torch.Tensor

Returns the buffer given by target if it exists, otherwise throws an error.

See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target.

Parameters: target – The fully-qualified string name of the buffer to look for. (See get_submodule for how to specify a fully-qualified string.)
Returns: The buffer referenced by target
Return type: torch.Tensor
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not a buffer

get_extra_state() → Any

Returns any extra state to include in the module’s state_dict. Implement this and a corresponding set_extra_state() for your module if you need to store extra state. This function is called when building the module’s state_dict().

Note that extra state should be pickleable to ensure working serialization of the state_dict. We only provide provide backwards compatibility guarantees for serializing Tensors; other objects may break backwards compatibility if their serialized pickled form changes.

Returns: Any extra state to store in the module’s state_dict
Return type: object

get_parameter(target: str) → torch.nn.parameter.Parameter

Returns the parameter given by target if it exists, otherwise throws an error.

See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target.

Parameters: target – The fully-qualified string name of the Parameter to look for. (See get_submodule for how to specify a fully-qualified string.)
Returns: The Parameter referenced by target
Return type: torch.nn.Parameter
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not an nn.Parameter

get_submodule(target: str) → torch.nn.modules.module.Module

Returns the submodule given by target if it exists, otherwise throws an error.

For example, let’s say you have an nn.Module A that looks like this:

A(
    (net_b): Module(
        (net_c): Module(
            (conv): Conv2d(16, 33, kernel_size=(3, 3), stride=(2, 2))
        )
        (linear): Linear(in_features=100, out_features=200, bias=True)
    )
)

(The diagram shows an nn.Module A. A has a nested submodule net_b, which itself has two submodules net_c and linear. net_c then has a submodule conv.)

To check whether or not we have the linear submodule, we would call get_submodule("net_b.linear"). To check whether we have the conv submodule, we would call get_submodule("net_b.net_c.conv").

The runtime of get_submodule is bounded by the degree of module nesting in target. A query against named_modules achieves the same result, but it is O(N) in the number of transitive modules. So, for a simple check to see if some submodule exists, get_submodule should always be used.

Parameters: target – The fully-qualified string name of the submodule to look for. (See above example for how to specify a fully-qualified string.)
Returns: The submodule referenced by target
Return type: torch.nn.Module
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not an nn.Module

half() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to half datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

ipu(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the IPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on IPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

load_state_dict(state_dict: Mapping[str, Any], strict: bool = True)

Copies parameters and buffers from state_dict into this module and its descendants. If strict is True, then the keys of state_dict must exactly match the keys returned by this module’s state_dict() function.

Parameters

state_dict (dict) – a dict containing parameters and persistent buffers.
strict (bool, optional) – whether to strictly enforce that the keys in state_dict match the keys returned by this module’s state_dict() function. Default: True

Returns

missing_keys is a list of str containing the missing keys
unexpected_keys is a list of str containing the unexpected keys

Return type

NamedTuple with missing_keys and unexpected_keys fields

Note

If a parameter or buffer is registered as None and its corresponding key exists in state_dict, load_state_dict() will raise a RuntimeError.

modules() → Iterator[torch.nn.modules.module.Module]

Returns an iterator over all modules in the network.

Yields: Module – a module in the network

Note

Duplicate modules are returned only once. In the following example, l will be returned only once.

Example:

>>> l = nn.Linear(2, 2)
>>> net = nn.Sequential(l, l)
>>> for idx, m in enumerate(net.modules()):
...     print(idx, '->', m)

0 -> Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
)
1 -> Linear(in_features=2, out_features=2, bias=True)

named_buffers(prefix: str = '', recurse: bool = True) → Iterator[Tuple[str, torch.Tensor]]

Returns an iterator over module buffers, yielding both the name of the buffer as well as the buffer itself.

Parameters

prefix (str) – prefix to prepend to all buffer names.
recurse (bool) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module.

Yields

(str, torch.Tensor) – Tuple containing the name and buffer

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, buf in self.named_buffers():
>>>    if name in ['running_var']:
>>>        print(buf.size())

named_children() → Iterator[Tuple[str, torch.nn.modules.module.Module]]

Returns an iterator over immediate children modules, yielding both the name of the module as well as the module itself.

Yields: (str, Module) – Tuple containing a name and child module

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, module in model.named_children():
>>>     if name in ['conv4', 'conv5']:
>>>         print(module)

named_modules(memo: Optional[Set[torch.nn.modules.module.Module]] = None, prefix: str = '', remove_duplicate: bool = True)

Returns an iterator over all modules in the network, yielding both the name of the module as well as the module itself.

Parameters

memo – a memo to store the set of modules already added to the result
prefix – a prefix that will be added to the name of the module
remove_duplicate – whether to remove the duplicated module instances in the result or not

Yields

(str, Module) – Tuple of name and module

Note

Duplicate modules are returned only once. In the following example, l will be returned only once.

Example:

>>> l = nn.Linear(2, 2)
>>> net = nn.Sequential(l, l)
>>> for idx, m in enumerate(net.named_modules()):
...     print(idx, '->', m)

0 -> ('', Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
))
1 -> ('0', Linear(in_features=2, out_features=2, bias=True))

named_parameters(prefix: str = '', recurse: bool = True) → Iterator[Tuple[str, torch.nn.parameter.Parameter]]

Returns an iterator over module parameters, yielding both the name of the parameter as well as the parameter itself.

Parameters

prefix (str) – prefix to prepend to all parameter names.
recurse (bool) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module.

Yields

(str, Parameter) – Tuple containing the name and parameter

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, param in self.named_parameters():
>>>    if name in ['bias']:
>>>        print(param.size())

parameters(recurse: bool = True) → Iterator[torch.nn.parameter.Parameter]

Returns an iterator over module parameters.

This is typically passed to an optimizer.

Parameters: recurse (bool) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module.
Yields: Parameter – module parameter

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for param in model.parameters():
>>>     print(type(param), param.size())
<class 'torch.Tensor'> (20L,)
<class 'torch.Tensor'> (20L, 1L, 5L, 5L)

register_backward_hook(hook: Callable[[torch.nn.modules.module.Module, Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[None, torch.Tensor]]) → torch.utils.hooks.RemovableHandle

Registers a backward hook on the module.

This function is deprecated in favor of register_full_backward_hook() and the behavior of this function will change in future versions.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_buffer(name: str, tensor: Optional[torch.Tensor], persistent: bool = True) → None

Adds a buffer to the module.

This is typically used to register a buffer that should not to be considered a model parameter. For example, BatchNorm’s running_mean is not a parameter, but is part of the module’s state. Buffers, by default, are persistent and will be saved alongside parameters. This behavior can be changed by setting persistent to False. The only difference between a persistent buffer and a non-persistent buffer is that the latter will not be a part of this module’s state_dict.

Buffers can be accessed as attributes using given names.

Parameters

name (str) – name of the buffer. The buffer can be accessed from this module using the given name
tensor (Tensor or None) – buffer to be registered. If None, then operations that run on buffers, such as cuda, are ignored. If None, the buffer is not included in the module’s state_dict.
persistent (bool) – whether the buffer is part of this module’s state_dict.

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> self.register_buffer('running_mean', torch.zeros(num_features))

register_forward_hook(hook: Callable[[...], None]) → torch.utils.hooks.RemovableHandle

Registers a forward hook on the module.

The hook will be called every time after forward() has computed an output. It should have the following signature:

hook(module, input, output) -> None or modified output

The input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the output. It can modify the input inplace but it will not have effect on forward since this is called after forward() is called.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_forward_pre_hook(hook: Callable[[...], None]) → torch.utils.hooks.RemovableHandle

Registers a forward pre-hook on the module.

The hook will be called every time before forward() is invoked. It should have the following signature:

hook(module, input) -> None or modified input

The input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the input. User can either return a tuple or a single modified value in the hook. We will wrap the value into a tuple if a single value is returned(unless that value is already a tuple).

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_full_backward_hook(hook: Callable[[torch.nn.modules.module.Module, Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[None, torch.Tensor]]) → torch.utils.hooks.RemovableHandle

Registers a backward hook on the module.

The hook will be called every time the gradients with respect to a module are computed, i.e. the hook will execute if and only if the gradients with respect to module outputs are computed. The hook should have the following signature:

hook(module, grad_input, grad_output) -> tuple(Tensor) or None

The grad_input and grad_output are tuples that contain the gradients with respect to the inputs and outputs respectively. The hook should not modify its arguments, but it can optionally return a new gradient with respect to the input that will be used in place of grad_input in subsequent computations. grad_input will only correspond to the inputs given as positional arguments and all kwarg arguments are ignored. Entries in grad_input and grad_output will be None for all non-Tensor arguments.

For technical reasons, when this hook is applied to a Module, its forward function will receive a view of each Tensor passed to the Module. Similarly the caller will receive a view of each Tensor returned by the Module’s forward function.

Warning

Modifying inputs or outputs inplace is not allowed when using backward hooks and will raise an error.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_load_state_dict_post_hook(hook)

Registers a post hook to be run after module’s load_state_dict is called.

It should have the following signature::: hook(module, incompatible_keys) -> None

The module argument is the current module that this hook is registered on, and the incompatible_keys argument is a NamedTuple consisting of attributes missing_keys and unexpected_keys. missing_keys is a list of str containing the missing keys and unexpected_keys is a list of str containing the unexpected keys.

The given incompatible_keys can be modified inplace if needed.

Note that the checks performed when calling load_state_dict() with strict=True are affected by modifications the hook makes to missing_keys or unexpected_keys, as expected. Additions to either set of keys will result in an error being thrown when strict=True, and clearning out both missing and unexpected keys will avoid an error.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_module(name: str, module: Optional[torch.nn.modules.module.Module]) → None: Alias for add_module().

register_parameter(name: str, param: Optional[torch.nn.parameter.Parameter]) → None

Adds a parameter to the module.

The parameter can be accessed as an attribute using given name.

Parameters

name (str) – name of the parameter. The parameter can be accessed from this module using the given name
param (Parameter or None) – parameter to be added to the module. If None, then operations that run on parameters, such as cuda, are ignored. If None, the parameter is not included in the module’s state_dict.

requires_grad_(requires_grad: bool = True) → torch.nn.modules.module.T

Change if autograd should record operations on parameters in this module.

This method sets the parameters’ requires_grad attributes in-place.

This method is helpful for freezing part of the module for finetuning or training parts of a model individually (e.g., GAN training).

See locally-disable-grad-doc for a comparison between .requires_grad_() and several similar mechanisms that may be confused with it.

Parameters: requires_grad (bool) – whether autograd should record operations on parameters in this module. Default: True.
Returns: self
Return type: Module

set_extra_state(state: Any)

This function is called from load_state_dict() to handle any extra state found within the state_dict. Implement this function and a corresponding get_extra_state() for your module if you need to store extra state within its state_dict.

Parameters: state (dict) – Extra state from the state_dict

share_memory() → torch.nn.modules.module.T: See torch.Tensor.share_memory_()

state_dict(*args, destination=None, prefix='', keep_vars=False)

Returns a dictionary containing references to the whole state of the module.

Both parameters and persistent buffers (e.g. running averages) are included. Keys are corresponding parameter and buffer names. Parameters and buffers set to None are not included.

Note

The returned object is a shallow copy. It contains references to the module’s parameters and buffers.

Warning

Currently state_dict() also accepts positional arguments for destination, prefix and keep_vars in order. However, this is being deprecated and keyword arguments will be enforced in future releases.

Warning

Please avoid the use of argument destination as it is not designed for end-users.

Parameters

destination (dict, optional) – If provided, the state of module will be updated into the dict and the same object is returned. Otherwise, an OrderedDict will be created and returned. Default: None.
prefix (str, optional) – a prefix added to parameter and buffer names to compose the keys in state_dict. Default: ''.
keep_vars (bool, optional) – by default the Tensor s returned in the state dict are detached from autograd. If it’s set to True, detaching will not be performed. Default: False.

Returns

a dictionary containing a whole state of the module

Return type

dict

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> module.state_dict().keys()
['bias', 'weight']

to(*args, **kwargs)

Moves and/or casts the parameters and buffers.

This can be called as

to(device=None, dtype=None, non_blocking=False)

to(dtype, non_blocking=False)

to(tensor, non_blocking=False)

to(memory_format=torch.channels_last)

Its signature is similar to torch.Tensor.to(), but only accepts floating point or complex dtypes. In addition, this method will only cast the floating point or complex parameters and buffers to dtype (if given). The integral parameters and buffers will be moved device, if that is given, but with dtypes unchanged. When non_blocking is set, it tries to convert/move asynchronously with respect to the host if possible, e.g., moving CPU Tensors with pinned memory to CUDA devices.

See below for examples.

Note

This method modifies the module in-place.

Parameters

device (torch.device) – the desired device of the parameters and buffers in this module
dtype (torch.dtype) – the desired floating point or complex dtype of the parameters and buffers in this module
tensor (torch.Tensor) – Tensor whose dtype and device are the desired dtype and device for all parameters and buffers in this module
memory_format (torch.memory_format) – the desired memory format for 4D parameters and buffers in this module (keyword only argument)

Returns

self

Return type

Module

Examples:

>>> # xdoctest: +IGNORE_WANT("non-deterministic")
>>> linear = nn.Linear(2, 2)
>>> linear.weight
Parameter containing:
tensor([[ 0.1913, -0.3420],
        [-0.5113, -0.2325]])
>>> linear.to(torch.double)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1913, -0.3420],
        [-0.5113, -0.2325]], dtype=torch.float64)
>>> # xdoctest: +REQUIRES(env:TORCH_DOCTEST_CUDA1)
>>> gpu1 = torch.device("cuda:1")
>>> linear.to(gpu1, dtype=torch.half, non_blocking=True)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1914, -0.3420],
        [-0.5112, -0.2324]], dtype=torch.float16, device='cuda:1')
>>> cpu = torch.device("cpu")
>>> linear.to(cpu)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1914, -0.3420],
        [-0.5112, -0.2324]], dtype=torch.float16)

>>> linear = nn.Linear(2, 2, bias=None).to(torch.cdouble)
>>> linear.weight
Parameter containing:
tensor([[ 0.3741+0.j,  0.2382+0.j],
        [ 0.5593+0.j, -0.4443+0.j]], dtype=torch.complex128)
>>> linear(torch.ones(3, 2, dtype=torch.cdouble))
tensor([[0.6122+0.j, 0.1150+0.j],
        [0.6122+0.j, 0.1150+0.j],
        [0.6122+0.j, 0.1150+0.j]], dtype=torch.complex128)

to_empty(*, device: Union[str, torch.device]) → torch.nn.modules.module.T

Moves the parameters and buffers to the specified device without copying storage.

Parameters: device (torch.device) – The desired device of the parameters and buffers in this module.
Returns: self
Return type: Module

train(mode: bool = True) → torch.nn.modules.module.T

Sets the module in training mode.

This has any effect only on certain modules. See documentations of particular modules for details of their behaviors in training/evaluation mode, if they are affected, e.g. Dropout, BatchNorm, etc.

Parameters: mode (bool) – whether to set training mode (True) or evaluation mode (False). Default: True.
Returns: self
Return type: Module

type(dst_type: Union[torch.dtype, str]) → torch.nn.modules.module.T

Casts all parameters and buffers to dst_type.

Note

This method modifies the module in-place.

Parameters: dst_type (type or string) – the desired type
Returns: self
Return type: Module

xpu(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the XPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on XPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

zero_grad(set_to_none: bool = False) → None

Sets gradients of all model parameters to zero. See similar function under torch.optim.Optimizer for more context.

Parameters: set_to_none (bool) – instead of setting to zero, set the grads to None. See torch.optim.Optimizer.zero_grad() for details.

Loss

class hierarchicalsoftmax.loss.HierarchicalSoftmaxLoss(root, **kwargs)

A module which sums the loss for each level of a hiearchical tree.

add_module(name: str, module: Optional[torch.nn.modules.module.Module]) → None

Adds a child module to the current module.

The module can be accessed as an attribute using the given name.

Parameters

name (str) – name of the child module. The child module can be accessed from this module using the given name
module (Module) – child module to be added to the module.

apply(fn: Callable[[torch.nn.modules.module.Module], None]) → torch.nn.modules.module.T

Applies fn recursively to every submodule (as returned by .children()) as well as self. Typical use includes initializing the parameters of a model (see also nn-init-doc).

Parameters: fn (Module -> None) – function to be applied to each submodule
Returns: self
Return type: Module

Example:

>>> @torch.no_grad()
>>> def init_weights(m):
>>>     print(m)
>>>     if type(m) == nn.Linear:
>>>         m.weight.fill_(1.0)
>>>         print(m.weight)
>>> net = nn.Sequential(nn.Linear(2, 2), nn.Linear(2, 2))
>>> net.apply(init_weights)
Linear(in_features=2, out_features=2, bias=True)
Parameter containing:
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
Linear(in_features=2, out_features=2, bias=True)
Parameter containing:
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
)

bfloat16() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to bfloat16 datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

buffers(recurse: bool = True) → Iterator[torch.Tensor]

Returns an iterator over module buffers.

Parameters: recurse (bool) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module.
Yields: torch.Tensor – module buffer

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for buf in model.buffers():
>>>     print(type(buf), buf.size())
<class 'torch.Tensor'> (20L,)
<class 'torch.Tensor'> (20L, 1L, 5L, 5L)

children() → Iterator[torch.nn.modules.module.Module]

Returns an iterator over immediate children modules.

Yields: Module – a child module

cpu() → torch.nn.modules.module.T

Moves all model parameters and buffers to the CPU.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

cuda(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the GPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on GPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

double() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to double datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

eval() → torch.nn.modules.module.T

Sets the module in evaluation mode.

This has any effect only on certain modules. See documentations of particular modules for details of their behaviors in training/evaluation mode, if they are affected, e.g. Dropout, BatchNorm, etc.

This is equivalent with self.train(False).

See locally-disable-grad-doc for a comparison between .eval() and several similar mechanisms that may be confused with it.

Returns: self
Return type: Module

extra_repr() → str

Set the extra representation of the module

To print customized extra information, you should re-implement this method in your own modules. Both single-line and multi-line strings are acceptable.

float() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to float datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

forward(batch_predictions: torch.Tensor, targets: torch.Tensor) → torch.Tensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

get_buffer(target: str) → torch.Tensor

Returns the buffer given by target if it exists, otherwise throws an error.

See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target.

Parameters: target – The fully-qualified string name of the buffer to look for. (See get_submodule for how to specify a fully-qualified string.)
Returns: The buffer referenced by target
Return type: torch.Tensor
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not a buffer

get_extra_state() → Any

Returns any extra state to include in the module’s state_dict. Implement this and a corresponding set_extra_state() for your module if you need to store extra state. This function is called when building the module’s state_dict().

Note that extra state should be pickleable to ensure working serialization of the state_dict. We only provide provide backwards compatibility guarantees for serializing Tensors; other objects may break backwards compatibility if their serialized pickled form changes.

Returns: Any extra state to store in the module’s state_dict
Return type: object

get_parameter(target: str) → torch.nn.parameter.Parameter

Returns the parameter given by target if it exists, otherwise throws an error.

See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target.

Parameters: target – The fully-qualified string name of the Parameter to look for. (See get_submodule for how to specify a fully-qualified string.)
Returns: The Parameter referenced by target
Return type: torch.nn.Parameter
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not an nn.Parameter

get_submodule(target: str) → torch.nn.modules.module.Module

Returns the submodule given by target if it exists, otherwise throws an error.

For example, let’s say you have an nn.Module A that looks like this:

A(
    (net_b): Module(
        (net_c): Module(
            (conv): Conv2d(16, 33, kernel_size=(3, 3), stride=(2, 2))
        )
        (linear): Linear(in_features=100, out_features=200, bias=True)
    )
)

(The diagram shows an nn.Module A. A has a nested submodule net_b, which itself has two submodules net_c and linear. net_c then has a submodule conv.)

To check whether or not we have the linear submodule, we would call get_submodule("net_b.linear"). To check whether we have the conv submodule, we would call get_submodule("net_b.net_c.conv").

The runtime of get_submodule is bounded by the degree of module nesting in target. A query against named_modules achieves the same result, but it is O(N) in the number of transitive modules. So, for a simple check to see if some submodule exists, get_submodule should always be used.

Parameters: target – The fully-qualified string name of the submodule to look for. (See above example for how to specify a fully-qualified string.)
Returns: The submodule referenced by target
Return type: torch.nn.Module
Raises: AttributeError – If the target string references an invalid path or resolves to something that is not an nn.Module

half() → torch.nn.modules.module.T

Casts all floating point parameters and buffers to half datatype.

Note

This method modifies the module in-place.

Returns: self
Return type: Module

ipu(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the IPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on IPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

load_state_dict(state_dict: Mapping[str, Any], strict: bool = True)

Copies parameters and buffers from state_dict into this module and its descendants. If strict is True, then the keys of state_dict must exactly match the keys returned by this module’s state_dict() function.

Parameters

state_dict (dict) – a dict containing parameters and persistent buffers.
strict (bool, optional) – whether to strictly enforce that the keys in state_dict match the keys returned by this module’s state_dict() function. Default: True

Returns

missing_keys is a list of str containing the missing keys
unexpected_keys is a list of str containing the unexpected keys

Return type

NamedTuple with missing_keys and unexpected_keys fields

Note

If a parameter or buffer is registered as None and its corresponding key exists in state_dict, load_state_dict() will raise a RuntimeError.

modules() → Iterator[torch.nn.modules.module.Module]

Returns an iterator over all modules in the network.

Yields: Module – a module in the network

Note

Duplicate modules are returned only once. In the following example, l will be returned only once.

Example:

>>> l = nn.Linear(2, 2)
>>> net = nn.Sequential(l, l)
>>> for idx, m in enumerate(net.modules()):
...     print(idx, '->', m)

0 -> Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
)
1 -> Linear(in_features=2, out_features=2, bias=True)

named_buffers(prefix: str = '', recurse: bool = True) → Iterator[Tuple[str, torch.Tensor]]

Returns an iterator over module buffers, yielding both the name of the buffer as well as the buffer itself.

Parameters

prefix (str) – prefix to prepend to all buffer names.
recurse (bool) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module.

Yields

(str, torch.Tensor) – Tuple containing the name and buffer

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, buf in self.named_buffers():
>>>    if name in ['running_var']:
>>>        print(buf.size())

named_children() → Iterator[Tuple[str, torch.nn.modules.module.Module]]

Returns an iterator over immediate children modules, yielding both the name of the module as well as the module itself.

Yields: (str, Module) – Tuple containing a name and child module

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, module in model.named_children():
>>>     if name in ['conv4', 'conv5']:
>>>         print(module)

named_modules(memo: Optional[Set[torch.nn.modules.module.Module]] = None, prefix: str = '', remove_duplicate: bool = True)

Returns an iterator over all modules in the network, yielding both the name of the module as well as the module itself.

Parameters

memo – a memo to store the set of modules already added to the result
prefix – a prefix that will be added to the name of the module
remove_duplicate – whether to remove the duplicated module instances in the result or not

Yields

(str, Module) – Tuple of name and module

Note

Duplicate modules are returned only once. In the following example, l will be returned only once.

Example:

>>> l = nn.Linear(2, 2)
>>> net = nn.Sequential(l, l)
>>> for idx, m in enumerate(net.named_modules()):
...     print(idx, '->', m)

0 -> ('', Sequential(
  (0): Linear(in_features=2, out_features=2, bias=True)
  (1): Linear(in_features=2, out_features=2, bias=True)
))
1 -> ('0', Linear(in_features=2, out_features=2, bias=True))

named_parameters(prefix: str = '', recurse: bool = True) → Iterator[Tuple[str, torch.nn.parameter.Parameter]]

Returns an iterator over module parameters, yielding both the name of the parameter as well as the parameter itself.

Parameters

prefix (str) – prefix to prepend to all parameter names.
recurse (bool) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module.

Yields

(str, Parameter) – Tuple containing the name and parameter

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for name, param in self.named_parameters():
>>>    if name in ['bias']:
>>>        print(param.size())

parameters(recurse: bool = True) → Iterator[torch.nn.parameter.Parameter]

Returns an iterator over module parameters.

This is typically passed to an optimizer.

Parameters: recurse (bool) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module.
Yields: Parameter – module parameter

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> for param in model.parameters():
>>>     print(type(param), param.size())
<class 'torch.Tensor'> (20L,)
<class 'torch.Tensor'> (20L, 1L, 5L, 5L)

register_backward_hook(hook: Callable[[torch.nn.modules.module.Module, Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[None, torch.Tensor]]) → torch.utils.hooks.RemovableHandle

Registers a backward hook on the module.

This function is deprecated in favor of register_full_backward_hook() and the behavior of this function will change in future versions.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_buffer(name: str, tensor: Optional[torch.Tensor], persistent: bool = True) → None

Adds a buffer to the module.

This is typically used to register a buffer that should not to be considered a model parameter. For example, BatchNorm’s running_mean is not a parameter, but is part of the module’s state. Buffers, by default, are persistent and will be saved alongside parameters. This behavior can be changed by setting persistent to False. The only difference between a persistent buffer and a non-persistent buffer is that the latter will not be a part of this module’s state_dict.

Buffers can be accessed as attributes using given names.

Parameters

name (str) – name of the buffer. The buffer can be accessed from this module using the given name
tensor (Tensor or None) – buffer to be registered. If None, then operations that run on buffers, such as cuda, are ignored. If None, the buffer is not included in the module’s state_dict.
persistent (bool) – whether the buffer is part of this module’s state_dict.

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> self.register_buffer('running_mean', torch.zeros(num_features))

register_forward_hook(hook: Callable[[...], None]) → torch.utils.hooks.RemovableHandle

Registers a forward hook on the module.

The hook will be called every time after forward() has computed an output. It should have the following signature:

hook(module, input, output) -> None or modified output

The input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the output. It can modify the input inplace but it will not have effect on forward since this is called after forward() is called.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_forward_pre_hook(hook: Callable[[...], None]) → torch.utils.hooks.RemovableHandle

Registers a forward pre-hook on the module.

The hook will be called every time before forward() is invoked. It should have the following signature:

hook(module, input) -> None or modified input

The input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the input. User can either return a tuple or a single modified value in the hook. We will wrap the value into a tuple if a single value is returned(unless that value is already a tuple).

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_full_backward_hook(hook: Callable[[torch.nn.modules.module.Module, Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[None, torch.Tensor]]) → torch.utils.hooks.RemovableHandle

Registers a backward hook on the module.

The hook will be called every time the gradients with respect to a module are computed, i.e. the hook will execute if and only if the gradients with respect to module outputs are computed. The hook should have the following signature:

hook(module, grad_input, grad_output) -> tuple(Tensor) or None

The grad_input and grad_output are tuples that contain the gradients with respect to the inputs and outputs respectively. The hook should not modify its arguments, but it can optionally return a new gradient with respect to the input that will be used in place of grad_input in subsequent computations. grad_input will only correspond to the inputs given as positional arguments and all kwarg arguments are ignored. Entries in grad_input and grad_output will be None for all non-Tensor arguments.

For technical reasons, when this hook is applied to a Module, its forward function will receive a view of each Tensor passed to the Module. Similarly the caller will receive a view of each Tensor returned by the Module’s forward function.

Warning

Modifying inputs or outputs inplace is not allowed when using backward hooks and will raise an error.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_load_state_dict_post_hook(hook)

Registers a post hook to be run after module’s load_state_dict is called.

It should have the following signature::: hook(module, incompatible_keys) -> None

The module argument is the current module that this hook is registered on, and the incompatible_keys argument is a NamedTuple consisting of attributes missing_keys and unexpected_keys. missing_keys is a list of str containing the missing keys and unexpected_keys is a list of str containing the unexpected keys.

The given incompatible_keys can be modified inplace if needed.

Note that the checks performed when calling load_state_dict() with strict=True are affected by modifications the hook makes to missing_keys or unexpected_keys, as expected. Additions to either set of keys will result in an error being thrown when strict=True, and clearning out both missing and unexpected keys will avoid an error.

Returns: a handle that can be used to remove the added hook by calling handle.remove()
Return type: torch.utils.hooks.RemovableHandle

register_module(name: str, module: Optional[torch.nn.modules.module.Module]) → None: Alias for add_module().

register_parameter(name: str, param: Optional[torch.nn.parameter.Parameter]) → None

Adds a parameter to the module.

The parameter can be accessed as an attribute using given name.

Parameters

name (str) – name of the parameter. The parameter can be accessed from this module using the given name
param (Parameter or None) – parameter to be added to the module. If None, then operations that run on parameters, such as cuda, are ignored. If None, the parameter is not included in the module’s state_dict.

requires_grad_(requires_grad: bool = True) → torch.nn.modules.module.T

Change if autograd should record operations on parameters in this module.

This method sets the parameters’ requires_grad attributes in-place.

This method is helpful for freezing part of the module for finetuning or training parts of a model individually (e.g., GAN training).

See locally-disable-grad-doc for a comparison between .requires_grad_() and several similar mechanisms that may be confused with it.

Parameters: requires_grad (bool) – whether autograd should record operations on parameters in this module. Default: True.
Returns: self
Return type: Module

set_extra_state(state: Any)

This function is called from load_state_dict() to handle any extra state found within the state_dict. Implement this function and a corresponding get_extra_state() for your module if you need to store extra state within its state_dict.

Parameters: state (dict) – Extra state from the state_dict

share_memory() → torch.nn.modules.module.T: See torch.Tensor.share_memory_()

state_dict(*args, destination=None, prefix='', keep_vars=False)

Returns a dictionary containing references to the whole state of the module.

Both parameters and persistent buffers (e.g. running averages) are included. Keys are corresponding parameter and buffer names. Parameters and buffers set to None are not included.

Note

The returned object is a shallow copy. It contains references to the module’s parameters and buffers.

Warning

Currently state_dict() also accepts positional arguments for destination, prefix and keep_vars in order. However, this is being deprecated and keyword arguments will be enforced in future releases.

Warning

Please avoid the use of argument destination as it is not designed for end-users.

Parameters

destination (dict, optional) – If provided, the state of module will be updated into the dict and the same object is returned. Otherwise, an OrderedDict will be created and returned. Default: None.
prefix (str, optional) – a prefix added to parameter and buffer names to compose the keys in state_dict. Default: ''.
keep_vars (bool, optional) – by default the Tensor s returned in the state dict are detached from autograd. If it’s set to True, detaching will not be performed. Default: False.

Returns

a dictionary containing a whole state of the module

Return type

dict

Example:

>>> # xdoctest: +SKIP("undefined vars")
>>> module.state_dict().keys()
['bias', 'weight']

to(*args, **kwargs)

Moves and/or casts the parameters and buffers.

This can be called as

to(device=None, dtype=None, non_blocking=False)

to(dtype, non_blocking=False)

to(tensor, non_blocking=False)

to(memory_format=torch.channels_last)

Its signature is similar to torch.Tensor.to(), but only accepts floating point or complex dtypes. In addition, this method will only cast the floating point or complex parameters and buffers to dtype (if given). The integral parameters and buffers will be moved device, if that is given, but with dtypes unchanged. When non_blocking is set, it tries to convert/move asynchronously with respect to the host if possible, e.g., moving CPU Tensors with pinned memory to CUDA devices.

See below for examples.

Note

This method modifies the module in-place.

Parameters

device (torch.device) – the desired device of the parameters and buffers in this module
dtype (torch.dtype) – the desired floating point or complex dtype of the parameters and buffers in this module
tensor (torch.Tensor) – Tensor whose dtype and device are the desired dtype and device for all parameters and buffers in this module
memory_format (torch.memory_format) – the desired memory format for 4D parameters and buffers in this module (keyword only argument)

Returns

self

Return type

Module

Examples:

>>> # xdoctest: +IGNORE_WANT("non-deterministic")
>>> linear = nn.Linear(2, 2)
>>> linear.weight
Parameter containing:
tensor([[ 0.1913, -0.3420],
        [-0.5113, -0.2325]])
>>> linear.to(torch.double)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1913, -0.3420],
        [-0.5113, -0.2325]], dtype=torch.float64)
>>> # xdoctest: +REQUIRES(env:TORCH_DOCTEST_CUDA1)
>>> gpu1 = torch.device("cuda:1")
>>> linear.to(gpu1, dtype=torch.half, non_blocking=True)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1914, -0.3420],
        [-0.5112, -0.2324]], dtype=torch.float16, device='cuda:1')
>>> cpu = torch.device("cpu")
>>> linear.to(cpu)
Linear(in_features=2, out_features=2, bias=True)
>>> linear.weight
Parameter containing:
tensor([[ 0.1914, -0.3420],
        [-0.5112, -0.2324]], dtype=torch.float16)

>>> linear = nn.Linear(2, 2, bias=None).to(torch.cdouble)
>>> linear.weight
Parameter containing:
tensor([[ 0.3741+0.j,  0.2382+0.j],
        [ 0.5593+0.j, -0.4443+0.j]], dtype=torch.complex128)
>>> linear(torch.ones(3, 2, dtype=torch.cdouble))
tensor([[0.6122+0.j, 0.1150+0.j],
        [0.6122+0.j, 0.1150+0.j],
        [0.6122+0.j, 0.1150+0.j]], dtype=torch.complex128)

to_empty(*, device: Union[str, torch.device]) → torch.nn.modules.module.T

Moves the parameters and buffers to the specified device without copying storage.

Parameters: device (torch.device) – The desired device of the parameters and buffers in this module.
Returns: self
Return type: Module

train(mode: bool = True) → torch.nn.modules.module.T

Sets the module in training mode.

This has any effect only on certain modules. See documentations of particular modules for details of their behaviors in training/evaluation mode, if they are affected, e.g. Dropout, BatchNorm, etc.

Parameters: mode (bool) – whether to set training mode (True) or evaluation mode (False). Default: True.
Returns: self
Return type: Module

type(dst_type: Union[torch.dtype, str]) → torch.nn.modules.module.T

Casts all parameters and buffers to dst_type.

Note

This method modifies the module in-place.

Parameters: dst_type (type or string) – the desired type
Returns: self
Return type: Module

xpu(device: Optional[Union[int, torch.device]] = None) → torch.nn.modules.module.T

Moves all model parameters and buffers to the XPU.

This also makes associated parameters and buffers different objects. So it should be called before constructing optimizer if the module will live on XPU while being optimized.

Note

This method modifies the module in-place.

Parameters: device (int, optional) – if specified, all parameters will be copied to that device
Returns: self
Return type: Module

zero_grad(set_to_none: bool = False) → None

Sets gradients of all model parameters to zero. See similar function under torch.optim.Optimizer for more context.

Parameters: set_to_none (bool) – instead of setting to zero, set the grads to None. See torch.optim.Optimizer.zero_grad() for details.

hierarchicalsoftmax.loss.focal_loss_with_smoothing(logits, label, weight=None, gamma=0.0, label_smoothing=0.0): Adapted from https://github.com/Kageshimasu/focal-loss-with-smoothing and https://github.com/clcarwin/focal_loss_pytorch

Inference

exception hierarchicalsoftmax.inference.ShapeError

Raised when the shape of a tensor is different to what is expected.

with_traceback(): Exception.with_traceback(tb) – set self.__traceback__ to tb and return self.

hierarchicalsoftmax.inference.greedy_prediction_node_ids(prediction_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode, max_depth: Optional[int] = None) → List[int]

Takes the prediction scores for a number of samples and converts it to a list of predictions of nodes in the tree.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

root (SoftmaxNode) – The root softmax node. Needs set_indexes to have been called.
prediction_tensor (torch.Tensor) – The predictions coming from the softmax layer. Shape (samples, root.layer_size)
max_depth (int, optional) – If set, then it only gives predictions at a maximum of this number of levels from the root.

Returns

A list of node IDs predicted for each sample.

Return type

List[int]

hierarchicalsoftmax.inference.greedy_predictions(prediction_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode, max_depth: Optional[int] = None, threshold: Optional[float] = None) → List[hierarchicalsoftmax.nodes.SoftmaxNode]

Takes the prediction scores for a number of samples and converts it to a list of predictions of nodes in the tree.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

prediction_tensor (torch.Tensor) – The output from the softmax layer. Shape (samples, root.layer_size) Works with raw scores or probabilities.
root (SoftmaxNode) – The root softmax node. Needs set_indexes to have been called.
prediction_tensor – The predictions coming from the softmax layer. Shape (samples, root.layer_size)
max_depth (int, optional) – If set, then it only gives predictions at a maximum of this number of levels from the root.
threshold (int, optional) – If set, then it only gives predictions where the value at the node is greater than this threshold. Designed for use with probabilities.

Returns

A list of nodes predicted for each sample.

Return type

List[nodes.SoftmaxNode]

hierarchicalsoftmax.inference.leaf_probabilities(prediction_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode) → torch.Tensor

hierarchicalsoftmax.inference.node_probabilities(prediction_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode) → torch.Tensor

hierarchicalsoftmax.inference.render_probabilities(root: hierarchicalsoftmax.nodes.SoftmaxNode, filepaths: Optional[List[pathlib.Path]] = None, prediction_color='red', non_prediction_color='gray', prediction_tensor: Optional[torch.Tensor] = None, probabilities: Optional[torch.Tensor] = None, predictions: Optional[List[hierarchicalsoftmax.nodes.SoftmaxNode]] = None, horizontal: bool = True, threshold: float = 0.005) → List[hierarchicalsoftmax.dotexporter.ThresholdDotExporter]

Renders the probabilities of each node in the tree as a graphviz graph.

See https://anytree.readthedocs.io/en/latest/_modules/anytree/exporter/dotexporter.html for more information.

Parameters

prediction_tensor (torch.Tensor) – The output activations from the softmax layer. Shape (samples, root.layer_size)
root (SoftmaxNode) – The root softmax node. Needs set_indexes to have been called.
filepaths (List[Path], optional) – Paths to locations where the files can be saved. Can have extension .dot or another format which can be interpreted by GraphViz such as .png or .svg. Defaults to None so that files are not saved.
prediction_color (str, optional) – The color for the greedy prediction nodes and edges. Defaults to “red”.
non_prediction_color (str, optional) – The color for the edges which weren’t predicted. Defaults to “gray”.

Returns

The list of rendered graphs.

Return type

List[DotExporter]

Metrics

hierarchicalsoftmax.metrics.greedy_accuracy(prediction_tensor, target_tensor, root, max_depth=None)

Gives the accuracy of predicting the target in a hierarchy tree.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

prediction_tensor (torch.Tensor) – A tensor with the raw scores for each node in the tree. Shape: (samples, root.layer_size)
target_tensor (torch.Tensor) – A tensor with the target node indexes. Shape: (samples,).
root (SoftmaxNode) – The root of the hierarchy tree.

Returns

The accuracy value (i.e. the number that are correct divided by the total number of samples)

Return type

float

hierarchicalsoftmax.metrics.greedy_accuracy_parent(prediction_tensor, target_tensor, root, max_depth=None)

Gives the accuracy of predicting the parent of the target in a hierarchy tree.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

prediction_tensor (torch.Tensor) – A tensor with the raw scores for each node in the tree. Shape: (samples, root.layer_size)
target_tensor (torch.Tensor) – A tensor with the target node indexes. Shape: (samples,).
root (SoftmaxNode) – The root of the hierarchy tree.

Returns

The accuracy value (i.e. the number that are correct divided by the total number of samples)

Return type

float

hierarchicalsoftmax.metrics.greedy_f1_score(prediction_tensor: torch.Tensor, target_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode, average: str = 'macro', max_depth=None) → float

Gives the f1 score of predicting the target i.e. a harmonic mean of the precision and recall.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

prediction_tensor (torch.Tensor) – A tensor with the raw scores for each node in the tree. Shape: (samples, root.layer_size)
target_tensor (torch.Tensor) – A tensor with the target node indexes. Shape: (samples,).
root (SoftmaxNode) – The root of the hierarchy tree.
average (str, optional) – The type of averaging over the different classes. Options are: ‘micro’, ‘macro’, ‘samples’, ‘weighted’, ‘binary’ or None. See https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html for details. Defaults to “macro”.

Returns

The f1 score

Return type

float

hierarchicalsoftmax.metrics.greedy_precision(prediction_tensor: torch.Tensor, target_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode, average: str = 'macro', max_depth=None) → float

Gives the precision score of predicting the target.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

prediction_tensor (torch.Tensor) – A tensor with the raw scores for each node in the tree. Shape: (samples, root.layer_size)
target_tensor (torch.Tensor) – A tensor with the target node indexes. Shape: (samples,).
root (SoftmaxNode) – The root of the hierarchy tree.
average (str, optional) – The type of averaging over the different classes. Options are: ‘micro’, ‘macro’, ‘samples’, ‘weighted’, ‘binary’ or None. See https://scikit-learn.org/stable/modules/generated/sklearn.metrics.precision_score.html for details. Defaults to “macro”.

Returns

The precision

Return type

float

hierarchicalsoftmax.metrics.greedy_recall(prediction_tensor: torch.Tensor, target_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode, average: str = 'macro', max_depth=None) → float

Gives the recall score of predicting the target.

Predictions use the greedy method which means that it chooses the greatest prediction score at each level of the tree.

Parameters

prediction_tensor (torch.Tensor) – A tensor with the raw scores for each node in the tree. Shape: (samples, root.layer_size)
target_tensor (torch.Tensor) – A tensor with the target node indexes. Shape: (samples,).
root (SoftmaxNode) – The root of the hierarchy tree.
average (str, optional) – The type of averaging over the different classes. Options are: ‘micro’, ‘macro’, ‘samples’, ‘weighted’, ‘binary’ or None. See https://scikit-learn.org/stable/modules/generated/sklearn.metrics.recall_score.html for details. Defaults to “macro”.

Returns

The recall

Return type

float

hierarchicalsoftmax.metrics.target_max_depth(target_tensor: torch.Tensor, root: hierarchicalsoftmax.nodes.SoftmaxNode, max_depth: int): Converts the target tensor to the max depth of the tree.