Losses#

class ivy.data_classes.container.losses._ContainerWithLosses(dict_in=None, queues=None, queue_load_sizes=None, container_combine_method='list_join', queue_timeout=None, print_limit=10, key_length_limit=None, print_indent=4, print_line_spacing=0, ivyh=None, default_key_color='green', keyword_color_dict=None, rebuild_child_containers=False, types_to_iteratively_nest=None, alphabetical_keys=True, dynamic_backend=None, build_callable=False, **kwargs)[source]#

Bases: ContainerBase

_abc_impl = <_abc._abc_data object>#
static _static_binary_cross_entropy(true, pred, /, *, from_logits=False, epsilon=0.0, reduction='mean', pos_weight=None, axis=None, key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container static method variant of ivy.binary_cross_entropy. This method simply wraps the function, and so the docstring for ivy.binary_cross_entropy also applies to this method with minimal changes.

Parameters:

  • true (Union[Container, Array, NativeArray]) – input array or container containing true labels.

  • pred (Union[Container, Array, NativeArray]) – input array or container containing Predicted labels.

  • from_logits (Union[bool, Container], default: False) – Whether pred is expected to be a logits tensor. By default, we assume that pred encodes a probability distribution.

  • epsilon (Union[float, Container], default: 0.0) – a float in [0.0, 1.0] specifying the amount of smoothing when calculating the loss. If epsilon is 0, no smoothing will be applied. Default: 0.

  • reduction (Union[str, Container], default: 'mean') – 'none': No reduction will be applied to the output. 'mean': The output will be averaged. 'sum': The output will be summed. Default: 'none'.

  • pos_weight (Optional[Union[Array, NativeArray, Container]], default: None) – a weight for positive examples. Must be an array with length equal to the number of classes.

  • axis (Optional[Union[int, Container]], default: None) – Axis along which to compute crossentropy.

  • key_chains (Optional[Union[List[str], Dict[str, str], Container]], default: None) – The key-chains to apply or not apply the method to. Default is None.

  • to_apply (Union[bool, Container], default: True) – If True, the method will be applied to key_chains, otherwise key_chains will be skipped. Default is True.

  • prune_unapplied (Union[bool, Container], default: False) – Whether to prune key_chains for which the function was not applied. Default is False.

  • map_sequences (Union[bool, Container], default: False) – Whether to also map method to sequences (lists, tuples). Default is False.

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The binary cross entropy between the given distributions.

Examples

With ivy.Container inputs:

>>> x = ivy.Container(a=ivy.array([1, 0, 0]),b=ivy.array([0, 0, 1]))
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = ivy.Container.static_binary_cross_entropy(x, y)
>>> print(z)
{
    a: ivy.array([0.511, 0.223, 0.357]),
    b: ivy.array([1.61, 0.223, 1.61])
}

With a mix of ivy.Array and ivy.Container inputs:

>>> x = ivy.array([1 , 1, 0])
>>> y = ivy.Container(a=ivy.array([0.7, 0.8, 0.2]),b=ivy.array([0.2, 0.6, 0.7]))
>>> z = ivy.Container.static_binary_cross_entropy(x, y)
>>> print(z)
{
    a: ivy.array([0.357, 0.223, 0.223]),
    b: ivy.array([1.61, 0.511, 1.2])
}
static _static_cross_entropy(true, pred, /, *, axis=-1, epsilon=1e-07, reduction='mean', key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container static method variant of ivy.cross_entropy. This method simply wraps the function, and so the docstring for ivy.cross_entropy also applies to this method with minimal changes.

Parameters:

  • true (Union[Container, Array, NativeArray]) – input array or container containing true labels.

  • pred (Union[Container, Array, NativeArray]) – input array or container containing the predicted labels.

  • axis (Union[int, Container], default: -1) – the axis along which to compute the cross-entropy. If axis is -1, the cross-entropy will be computed along the last dimension. Default: -1.

  • epsilon (Union[float, Container], default: 1e-07) – a float in [0.0, 1.0] specifying the amount of smoothing when calculating the loss. If epsilon is 0, no smoothing will be applied. Default: 1e-7.

  • key_chains (Optional[Union[List[str], Dict[str, str], Container]], default: None) – The key-chains to apply or not apply the method to. Default is None.

  • to_apply (Union[bool, Container], default: True) – If True, the method will be applied to key_chains, otherwise key_chains will be skipped. Default is True.

  • prune_unapplied (Union[bool, Container], default: False) – Whether to prune key_chains for which the function was not applied. Default is False.

  • map_sequences (Union[bool, Container], default: False) – Whether to also map method to sequences (lists, tuples). Default is False.

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The cross-entropy loss between the given distributions.

Examples

With ivy.Container inputs:

>>> x = ivy.Container(a=ivy.array([0, 0, 1]), b=ivy.array([1, 1, 0]))
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = ivy.Container.static_cross_entropy(x, y)
>>> print(z)
{
    a: ivy.array(1.20397282),
    b: ivy.array(1.83258148)
}

With a mix of ivy.Array and ivy.Container inputs:

>>> x = ivy.array([0, 0, 1])
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = ivy.Container.static_cross_entropy(x, y)
>>> print(z)
{
    a: ivy.array(1.20397282),
    b: ivy.array(1.60943794)
}
static _static_sparse_cross_entropy(true, pred, /, *, axis=-1, epsilon=1e-07, reduction='mean', key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container static method variant of ivy.sparse_cross_entropy. This method simply wraps the function, and so the docstring for ivy.sparse_cross_entropy also applies to this method with minimal changes.

Parameters:

  • true (Union[Container, Array, NativeArray]) – input array or container containing the true labels as logits.

  • pred (Union[Container, Array, NativeArray]) – input array or container containing the predicted labels as logits.

  • axis (Union[int, Container], default: -1) – the axis along which to compute the cross-entropy. If axis is -1, the cross-entropy will be computed along the last dimension. Default: -1. epsilon a float in [0.0, 1.0] specifying the amount of smoothing when calculating the loss. If epsilon is 0, no smoothing will be applied. Default: 1e-7.

  • key_chains (Optional[Union[List[str], Dict[str, str], Container]], default: None) – The key-chains to apply or not apply the method to. Default is None.

  • to_apply (Union[bool, Container], default: True) – If True, the method will be applied to key_chains, otherwise key_chains will be skipped. Default is True.

  • prune_unapplied (Union[bool, Container], default: False) – Whether to prune key_chains for which the function was not applied. Default is False.

  • map_sequences (Union[bool, Container], default: False) – Whether to also map method to sequences (lists, tuples). Default is False.

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The sparse cross-entropy loss between the given distributions.

Examples

With ivy.Container inputs:

>>> x = ivy.Container(a=ivy.array([1, 0, 0]),b=ivy.array([0, 0, 1]))
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = ivy.Container.static_sparse_cross_entropy(x, y)
>>> print(z)
{
    a: ivy.array([1.61, 0.511, 0.511]),
    b: ivy.array([0.223, 0.223, 1.61])
}

With a mix of ivy.Array and ivy.Container inputs:

>>> x = ivy.array([1 , 1, 0])
>>> y = ivy.Container(a=ivy.array([0.7, 0.8, 0.2]),b=ivy.array([0.2, 0.6, 0.7]))
>>> z = ivy.Container.static_sparse_cross_entropy(x, y)
>>> print(z)
{
    a: ivy.array([0.223, 0.223, 0.357]),
    b: ivy.array([0.511, 0.511, 1.61])
}
binary_cross_entropy(pred, /, *, from_logits=False, epsilon=0.0, reduction='mean', pos_weight=None, axis=None, key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container instance method variant of ivy.binary_cross_entropy. This method simply wraps the function, and so the docstring for ivy.binary_cross_entropy also applies to this method with minimal changes.

Parameters:

  • self (Container) – input container containing true labels.

  • pred (Union[Container, Array, NativeArray]) –

    input array or container containing Predicted labels. from_logits

    Whether pred is expected to be a logits tensor. By default, we assume that pred encodes a probability distribution.

  • epsilon (Union[float, Container], default: 0.0) – a float in [0.0, 1.0] specifying the amount of smoothing when calculating the loss. If epsilon is 0, no smoothing will be applied. Default: 0.

  • reduction (Union[str, Container], default: 'mean') – 'none': No reduction will be applied to the output. 'mean': The output will be averaged. 'sum': The output will be summed. Default: 'none'.

  • pos_weight (Optional[Union[Array, NativeArray, Container]], default: None) – a weight for positive examples. Must be an array with length equal to the number of classes.

  • axis (Optional[Union[int, Container]], default: None) – Axis along which to compute crossentropy.

  • key_chains (Optional[Union[List[str], Dict[str, str], Container]], default: None) – The key-chains to apply or not apply the method to. Default is None.

  • to_apply (Union[bool, Container], default: True) – If True, the method will be applied to key_chains, otherwise key_chains will be skipped. Default is True.

  • prune_unapplied (Union[bool, Container], default: False) – Whether to prune key_chains for which the function was not applied. Default is False.

  • map_sequences (Union[bool, Container], default: False) – Whether to also map method to sequences (lists, tuples). Default is False.

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The binary cross entropy between the given distributions.

Examples

>>> x = ivy.Container(a=ivy.array([1, 0, 0]),b=ivy.array([0, 0, 1]))
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = x.binary_cross_entropy(y)
>>> print(z)
{
    a: ivy.array([0.511, 0.223, 0.357]),
    b: ivy.array([1.61, 0.223, 1.61])
}
cross_entropy(pred, /, *, axis=-1, epsilon=1e-07, reduction='mean', key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container instance method variant of ivy.cross_entropy. This method simply wraps the function, and so the docstring for ivy.cross_entropy also applies to this method with minimal changes.

Parameters:

  • self (Container) – input container containing true labels.

  • pred (Union[Container, Array, NativeArray]) – input array or container containing the predicted labels.

  • axis (Union[int, Container], default: -1) – the axis along which to compute the cross-entropy. If axis is -1, the cross-entropy will be computed along the last dimension. Default: -1.

  • epsilon (Union[float, Container], default: 1e-07) – a float in [0.0, 1.0] specifying the amount of smoothing when calculating the loss. If epsilon is 0, no smoothing will be applied. Default: 1e-7.

  • key_chains (Optional[Union[List[str], Dict[str, str], Container]], default: None) – The key-chains to apply or not apply the method to. Default is None.

  • to_apply (Union[bool, Container], default: True) – If True, the method will be applied to key_chains, otherwise key_chains will be skipped. Default is True.

  • prune_unapplied (Union[bool, Container], default: False) – Whether to prune key_chains for which the function was not applied. Default is False.

  • map_sequences (Union[bool, Container], default: False) – Whether to also map method to sequences (lists, tuples). Default is False.

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The cross-entropy loss between the given distributions.

Examples

>>> x = ivy.Container(a=ivy.array([1, 0, 0]),b=ivy.array([0, 0, 1]))
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = x.cross_entropy(y)
>>> print(z)
{
    a:ivy.array(0.5108256),
    b:ivy.array(1.609438)
}
sparse_cross_entropy(pred, /, *, axis=-1, epsilon=1e-07, reduction='mean', key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container instance method variant of ivy.sparse_cross_entropy. This method simply wraps the function, and so the docstring for ivy.sparse_cross_entropy also applies to this method with minimal changes.

Parameters:

  • self (Container) – input container containing the true labels as logits.

  • pred (Union[Container, Array, NativeArray]) – input array or container containing the predicted labels as logits.

  • axis (Union[int, Container], default: -1) – the axis along which to compute the cross-entropy. If axis is -1, the cross-entropy will be computed along the last dimension. Default: -1. epsilon a float in [0.0, 1.0] specifying the amount of smoothing when calculating the loss. If epsilon is 0, no smoothing will be applied. Default: 1e-7.

  • key_chains (Optional[Union[List[str], Dict[str, str], Container]], default: None) – The key-chains to apply or not apply the method to. Default is None.

  • to_apply (Union[bool, Container], default: True) – If True, the method will be applied to key_chains, otherwise key_chains will be skipped. Default is True.

  • prune_unapplied (Union[bool, Container], default: False) – Whether to prune key_chains for which the function was not applied. Default is False.

  • map_sequences (Union[bool, Container], default: False) – Whether to also map method to sequences (lists, tuples). Default is False.

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The sparse cross-entropy loss between the given distributions.

Examples

>>> x = ivy.Container(a=ivy.array([1, 0, 0]),b=ivy.array([0, 0, 1]))
>>> y = ivy.Container(a=ivy.array([0.6, 0.2, 0.3]),b=ivy.array([0.8, 0.2, 0.2]))
>>> z = x.sparse_cross_entropy(y)
>>> print(z)
{
    a: ivy.array([1.61, 0.511, 0.511]),
    b: ivy.array([0.223, 0.223, 1.61])
}