PermuteTransform

class torchrl.envs.transforms.PermuteTransform(dims, in_keys=None, out_keys=None, in_keys_inv=None, out_keys_inv=None)

排列变换。

沿所需维度排列输入张量。排列必须沿特征维度（而非批次维度）提供。

参数:

• dims (list of int) – 维度的排列顺序。必须是 dims [-(len(dims)), ..., -1] 的重新排序。

• in_keys (list of NestedKeys) – 输入条目（读取）。

• out_keys (list of NestedKeys) – 输入条目（写入）。如果未提供，则默认为 in_keys。

• in_keys_inv (list of NestedKeys) – inv() 调用期间的输入条目（读取）。

• out_keys_inv (list of NestedKeys) – inv() 调用期间的输入条目（写入）。如果未提供，则默认为 in_keys_in。

示例

>>> from torchrl.envs.libs.gym import GymEnv
>>> base_env = GymEnv("ALE/Pong-v5")
>>> base_env.rollout(2)
TensorDict(
    fields={
        action: Tensor(shape=torch.Size([2, 6]), device=cpu, dtype=torch.int64, is_shared=False),
        done: Tensor(shape=torch.Size([2, 1]), device=cpu, dtype=torch.bool, is_shared=False),
        next: TensorDict(
            fields={
                done: Tensor(shape=torch.Size([2, 1]), device=cpu, dtype=torch.bool, is_shared=False),
                pixels: Tensor(shape=torch.Size([2, 210, 160, 3]), device=cpu, dtype=torch.uint8, is_shared=False),
                reward: Tensor(shape=torch.Size([2, 1]), device=cpu, dtype=torch.float32, is_shared=False)},
            batch_size=torch.Size([2]),
            device=cpu,
            is_shared=False),
        pixels: Tensor(shape=torch.Size([2, 210, 160, 3]), device=cpu, dtype=torch.uint8, is_shared=False)},
    batch_size=torch.Size([2]),
    device=cpu,
    is_shared=False)
>>> env = TransformedEnv(base_env, PermuteTransform((-1, -3, -2), in_keys=["pixels"]))
>>> env.rollout(2)  # channels are at the end
TensorDict(
    fields={
        action: Tensor(shape=torch.Size([2, 6]), device=cpu, dtype=torch.int64, is_shared=False),
        done: Tensor(shape=torch.Size([2, 1]), device=cpu, dtype=torch.bool, is_shared=False),
        next: TensorDict(
            fields={
                done: Tensor(shape=torch.Size([2, 1]), device=cpu, dtype=torch.bool, is_shared=False),
                pixels: Tensor(shape=torch.Size([2, 3, 210, 160]), device=cpu, dtype=torch.uint8, is_shared=False),
                reward: Tensor(shape=torch.Size([2, 1]), device=cpu, dtype=torch.float32, is_shared=False)},
            batch_size=torch.Size([2]),
            device=cpu,
            is_shared=False),
        pixels: Tensor(shape=torch.Size([2, 3, 210, 160]), device=cpu, dtype=torch.uint8, is_shared=False)},
    batch_size=torch.Size([2]),
    device=cpu,
    is_shared=False)

transform_input_spec(input_spec)

转换输入规范，使结果规范与变换映射匹配。

参数:

input_spec (TensorSpec) – 变换前的规范

返回:

变换后预期的规范

transform_observation_spec(observation_spec: TensorSpec) → TensorSpec

转换观测规范，使结果规范与变换映射匹配。

参数:

observation_spec (TensorSpec) – 变换前的规范

返回:

变换后预期的规范