StackedComposite

class torchrl.data.StackedComposite(*args, **kwargs)

A lazy representation of a stack of composite specs.

Stacks composite specs together along one dimension. When random samples are drawn, a LazyStackedTensorDict is returned.

Indexing is allowed but only along the stack dimension.

This class is aimed to be used in multi-task and multi-agent settings, where heterogeneous specs may occur (same semantic but different shape).

assert_is_in(value: Tensor) → None

Asserts whether a tensor belongs to the box, and raises an exception otherwise.

Parameters:

value (torch.Tensor) – value to be checked.

clear_device_()

Clears the device of the Composite.

clone() → T

Creates a copy of the TensorSpec.

contains(item: torch.Tensor | tensordict.base.TensorDictBase) → bool

If the value val could have been generated by the TensorSpec, returns True, otherwise False.

See is_in() for more information.

cpu()

Casts the TensorSpec to ‘cpu’ device.

cuda(device=None)

Casts the TensorSpec to ‘cuda’ device.

property device: Union[device, str, int]

The device of the spec.

Only Composite specs can have a None device. All leaves must have a non-null device.

empty()

Create a spec like self, but with no entries.

encode(vals: Dict[str, Any], ignore_device: bool = False) → Dict[str, Tensor]

Encodes a value given the specified spec, and return the corresponding tensor.

This method is to be used in environments that return a value (eg, a numpy array) that can be easily mapped to the TorchRL required domain. If the value is already a tensor, the spec will not change its value and return it as-is.

Parameters:

val (np.ndarray or torch.Tensor) – value to be encoded as tensor.

Keyword Arguments:

ignore_device (bool, optional) – if True, the spec device will be ignored. This is used to group tensor casting within a call to TensorDict(..., device="cuda") which is faster.

Returns:

torch.Tensor matching the required tensor specs.

enumerate() → TensorDictBase

Returns all the samples that can be obtained from the TensorSpec.

The samples will be stacked along the first dimension.

This method is only implemented for discrete specs.

expand(*shape)

Returns a new Spec with the expanded shape.

Parameters:

*shape (tuple or iterable of int) – the new shape of the Spec. Must be broadcastable with the current shape: its length must be at least as long as the current shape length, and its last values must be compliant too; ie they can only differ from it if the current dimension is a singleton.

flatten(start_dim: int, end_dim: int) → T

Flattens a TensorSpec.

Check flatten() for more information on this method.

get(item, default=_NoDefault.ZERO)

Gets an item from the Composite.

If the item is absent, a default value can be passed.

classmethod implements_for_spec(torch_function: Callable) → Callable

Register a torch function override for TensorSpec.

abstract index(index: Union[int, Tensor, ndarray, slice, List], tensor_to_index: torch.Tensor | tensordict.base.TensorDictBase) → torch.Tensor | tensordict.base.TensorDictBase

Indexes the input tensor.

Parameters:

• index (int, torch.Tensor, slice or list) – index of the tensor

• tensor_to_index – tensor to be indexed

Returns:

indexed tensor

is_empty()

Whether the composite spec contains specs or not.

is_in(value) → bool

If the value val could have been generated by the TensorSpec, returns True, otherwise False.

More precisely, the is_in methods checks that the value val is within the limits defined by the space attribute (the box), and that the dtype, device, shape potentially other metadata match those of the spec. If any of these checks fails, the is_in method will return False.

Parameters:

val (torch.Tensor) – value to be checked.

Returns:

boolean indicating if values belongs to the TensorSpec box.

items(include_nested: bool = False, leaves_only: bool = False, *, is_leaf: Optional[Callable[[type], bool]] = None) → _CompositeSpecItemsView

Items of the Composite.

Parameters:

• include_nested (bool, optional) – if False, the returned keys will not be nested. They will represent only the immediate children of the root, and not the whole nested sequence, i.e. a Composite(next=Composite(obs=None)) will lead to the keys ["next"]. Default is ``False`, i.e. nested keys will not be returned.

• leaves_only (bool, optional) – if False, the values returned will contain every level of nesting, i.e. a Composite(next=Composite(obs=None)) will lead to the keys ["next", ("next", "obs")]. Default is False.

Keyword Arguments:

is_leaf (callable, optional) – reads a type and returns a boolean indicating if that type should be seen as a leaf. By default, all non-Composite nodes are considered as leaves.

keys(include_nested: bool = False, leaves_only: bool = False, *, is_leaf: Optional[Callable[[type], bool]] = None) → _CompositeSpecKeysView

Keys of the Composite.

The keys argument reflect those of tensordict.TensorDict.

Parameters:

• include_nested (bool, optional) – if False, the returned keys will not be nested. They will represent only the immediate children of the root, and not the whole nested sequence, i.e. a Composite(next=Composite(obs=None)) will lead to the keys ["next"]. Default is ``False`, i.e. nested keys will not be returned.

• leaves_only (bool, optional) – if False, the values returned will contain every level of nesting, i.e. a Composite(next=Composite(obs=None)) will lead to the keys ["next", ("next", "obs")]. Default is False.

Keyword Arguments:

is_leaf (callable, optional) – reads a type and returns a boolean indicating if that type should be seen as a leaf. By default, all non-Composite nodes are considered as leaves.

lock_(recurse=False)

Locks the Composite and prevents modification of its content.

This is only a first-level lock, unless specified otherwise through the recurse arg.

Leaf specs can always be modified in place, but they cannot be replaced in their Composite parent.

Examples

>>> shape = [3, 4, 5]
>>> spec = Composite(
...         a=Composite(
...         b=Composite(shape=shape[:3], device="cpu"), shape=shape[:2]
...     ),
...     shape=shape[:1],
... )
>>> spec["a"] = spec["a"].clone()
>>> recurse = False
>>> spec.lock_(recurse=recurse)
>>> try:
...     spec["a"] = spec["a"].clone()
... except RuntimeError:
...     print("failed!")
failed!
>>> try:
...     spec["a", "b"] = spec["a", "b"].clone()
...     print("succeeded!")
... except RuntimeError:
...     print("failed!")
succeeded!
>>> recurse = True
>>> spec.lock_(recurse=recurse)
>>> try:
...     spec["a", "b"] = spec["a", "b"].clone()
...     print("succeeded!")
... except RuntimeError:
...     print("failed!")
failed!

make_neg_dim(dim: int)

Converts a specific dimension to -1.

property ndim

Number of dimensions of the spec shape.

Shortcut for len(spec.shape).

ndimension()

Number of dimensions of the spec shape.

Shortcut for len(spec.shape).

one(shape: Optional[Size] = None) → TensorDictBase

Returns a one-filled tensor in the box.

Note

Even though there is no guarantee that 1 belongs to the spec domain, this method will not raise an exception when this condition is violated. The primary use case of one is to generate empty data buffers, not meaningful data.

Parameters:

shape (torch.Size) – shape of the one-tensor

Returns:

a one-filled tensor sampled in the TensorSpec box.

ones(shape: Optional[Size] = None) → torch.Tensor | tensordict.base.TensorDictBase

Proxy to one().

project(val: TensorDictBase) → TensorDictBase

If the input tensor is not in the TensorSpec box, it maps it back to it given some defined heuristic.

Parameters:

val (torch.Tensor) – tensor to be mapped to the box.

Returns:

a torch.Tensor belonging to the TensorSpec box.

rand(shape: Optional[Size] = None) → TensorDictBase

Returns a random tensor in the space defined by the spec.

The sampling will be done uniformly over the space, unless the box is unbounded in which case normal values will be drawn.

Parameters:

shape (torch.Size) – shape of the random tensor

Returns:

a random tensor sampled in the TensorSpec box.

reshape(*shape) → T

Reshapes a TensorSpec.

Check reshape() for more information on this method.

sample(shape: Optional[Size] = None) → torch.Tensor | tensordict.base.TensorDictBase

Returns a random tensor in the space defined by the spec.

See rand() for details.

squeeze(dim: Optional[int] = None)

Returns a new Spec with all the dimensions of size 1 removed.

When dim is given, a squeeze operation is done only in that dimension.

Parameters:

dim (int or None) – the dimension to apply the squeeze operation to

to(dest: Union[dtype, device, str, int]) → T

Casts a TensorSpec to a device or a dtype.

Returns the same spec if no change is made.

to_numpy(val: TensorDict, safe: Optional[bool] = None) → dict

Returns the np.ndarray correspondent of an input tensor.

This is intended to be the inverse operation of encode().

Parameters:

• val (torch.Tensor) – tensor to be transformed_in to numpy.

• safe (bool) – boolean value indicating whether a check should be performed on the value against the domain of the spec. Defaults to the value of the CHECK_SPEC_ENCODE environment variable.

Returns:

a np.ndarray.

type_check(value: Union[Tensor, TensorDictBase], selected_keys: Optional[Union[NestedKey, Sequence[NestedKey]]] = None)

Checks the input value dtype against the TensorSpec dtype and raises an exception if they don’t match.

Parameters:

• value (torch.Tensor) – tensor whose dtype has to be checked.

• key (str, optional) – if the TensorSpec has keys, the value dtype will be checked against the spec pointed by the indicated key.

unflatten(dim: int, sizes: Tuple[int]) → T

Unflattens a TensorSpec.

Check unflatten() for more information on this method.

unlock_(recurse=False)

Unlocks the Composite and allows modification of its content.

This is only a first-level lock modification, unless specified otherwise through the recurse arg.

unsqueeze(dim: int)

Returns a new Spec with one more singleton dimension (at the position indicated by dim).

Parameters:

dim (int or None) – the dimension to apply the unsqueeze operation to.

values(include_nested: bool = False, leaves_only: bool = False, *, is_leaf: Optional[Callable[[type], bool]] = None) → _CompositeSpecValuesView

Values of the Composite.

Parameters:

• include_nested (bool, optional) – if False, the returned keys will not be nested. They will represent only the immediate children of the root, and not the whole nested sequence, i.e. a Composite(next=Composite(obs=None)) will lead to the keys ["next"]. Default is ``False`, i.e. nested keys will not be returned.

• leaves_only (bool, optional) – if False, the values returned will contain every level of nesting, i.e. a Composite(next=Composite(obs=None)) will lead to the keys ["next", ("next", "obs")]. Default is False.

Keyword Arguments:

is_leaf (callable, optional) – reads a type and returns a boolean indicating if that type should be seen as a leaf. By default, all non-Composite nodes are considered as leaves.

view(*shape) → T

Reshapes a TensorSpec.

Check reshape() for more information on this method.

zero(shape: Optional[Size] = None) → TensorDictBase

Returns a zero-filled tensor in the box.

Note

Even though there is no guarantee that 0 belongs to the spec domain, this method will not raise an exception when this condition is violated. The primary use case of zero is to generate empty data buffers, not meaningful data.

Parameters:

shape (torch.Size) – shape of the zero-tensor

Returns:

a zero-filled tensor sampled in the TensorSpec box.

zeros(shape: Optional[Size] = None) → torch.Tensor | tensordict.base.TensorDictBase

Proxy to zero().