Transform

class torchrl.envs.transforms.Transform(in_keys: Sequence[NestedKey] = None, out_keys: Sequence[NestedKey] | None = None, in_keys_inv: Sequence[NestedKey] | None = None, out_keys_inv: Sequence[NestedKey] | None = None)

Environment transform parent class.

In principle, a transform receives a tensordict as input and returns ( the same or another) tensordict as output, where a series of values have been modified or created with a new key. When instantiating a new transform, the keys that are to be read from are passed to the constructor via the keys argument.

Transforms are to be combined with their target environments with the TransformedEnv class, which takes as arguments an EnvBase instance and a transform. If multiple transforms are to be used, they can be concatenated using the Compose class. A transform can be stateless or stateful (e.g. CatTransform). Because of this, Transforms support the reset operation, which should reset the transform to its initial state (such that successive trajectories are kept independent).

Notably, Transform subclasses take care of transforming the affected specs from an environment: when querying transformed_env.observation_spec, the resulting objects will describe the specs of the transformed_in tensors.

property container

Returns the env containing the transform.

Examples

>>> from torchrl.envs import TransformedEnv, Compose, RewardSum, StepCounter
>>> from torchrl.envs.libs.gym import GymEnv
>>> env = TransformedEnv(GymEnv("Pendulum-v1"), Compose(RewardSum(), StepCounter()))
>>> env.transform[0].container is env
True

forward(tensordict: TensorDictBase) → TensorDictBase

Reads the input tensordict, and for the selected keys, applies the transform.

property parent: Optional[EnvBase]

Returns the parent env of the transform.

The parent env is the env that contains all the transforms up until the current one.

Examples

>>> from torchrl.envs import TransformedEnv, Compose, RewardSum, StepCounter
>>> from torchrl.envs.libs.gym import GymEnv
>>> env = TransformedEnv(GymEnv("Pendulum-v1"), Compose(RewardSum(), StepCounter()))
>>> env.transform[1].parent
TransformedEnv(
    env=GymEnv(env=Pendulum-v1, batch_size=torch.Size([]), device=cpu),
    transform=Compose(
            RewardSum(keys=['reward'])))

to(*args, **kwargs)

Move and/or cast 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)

transform_done_spec(done_spec: TensorSpec) → TensorSpec

Transforms the done spec such that the resulting spec matches transform mapping.

Parameters:

done_spec (TensorSpec) – spec before the transform

Returns:

expected spec after the transform

transform_env_batch_size(batch_size: Size)

Transforms the batch-size of the parent env.

transform_env_device(device: device)

Transforms the device of the parent env.

transform_input_spec(input_spec: TensorSpec) → TensorSpec

Transforms the input spec such that the resulting spec matches transform mapping.

Parameters:

input_spec (TensorSpec) – spec before the transform

Returns:

expected spec after the transform

transform_observation_spec(observation_spec: TensorSpec) → TensorSpec

Transforms the observation spec such that the resulting spec matches transform mapping.

Parameters:

observation_spec (TensorSpec) – spec before the transform

Returns:

expected spec after the transform

transform_output_spec(output_spec: CompositeSpec) → CompositeSpec

Transforms the output spec such that the resulting spec matches transform mapping.

This method should generally be left untouched. Changes should be implemented using transform_observation_spec(), transform_reward_spec() and transformfull_done_spec(). :param output_spec: spec before the transform :type output_spec: TensorSpec

Returns:

expected spec after the transform

transform_reward_spec(reward_spec: TensorSpec) → TensorSpec

Transforms the reward spec such that the resulting spec matches transform mapping.

Parameters:

reward_spec (TensorSpec) – spec before the transform

Returns:

expected spec after the transform