torchrl.utils package

Submodules

torchrl.utils.schedule module

class torchrl.utils.schedule.ExpDecaySchedule(start=1.0, end=0.01, num_steps=1000)

Bases: torchrl.utils.schedule.Schedule

step()

Step the wrapped_value

property wrapped_value

Access wrapped value without side-effect

class torchrl.utils.schedule.LinearSchedule(min_val=0.0, max_val=1.0, num_steps=1000, invert=False)

Bases: torchrl.utils.schedule.Schedule

step()

Step the wrapped_value

property wrapped_value

Access wrapped value without side-effect

class torchrl.utils.schedule.Schedule

Bases: object

abstract step()

Step the wrapped_value

property value

Step schedule and return the old value

property wrapped_value

Access wrapped value without side-effect

torchrl.utils.storage module

class torchrl.utils.storage.TensorTupleDataset(size: int = -1, x: Optional[List] = None)

Bases: torch.utils.data.dataset.Dataset

Store vectorized tuples of tensors

extend(*tensor_list: List[torch.Tensor])

truncate()

class torchrl.utils.storage.Transition(obs, action, reward, next_obs, done)

Bases: tuple

property action

Alias for field number 1

property done

Alias for field number 4

property next_obs

Alias for field number 3

property obs

Alias for field number 0

property reward

Alias for field number 2

class torchrl.utils.storage.TransitionTupleDataset(size: int = -1, x: Optional[List] = None)

Bases: torchrl.utils.storage.TensorTupleDataset

extend(transition_list: List[torchrl.utils.storage.Transition])

torchrl.utils.storage.truncated_cat(a, b, maxsize=-1)

torchrl.utils.test_schedule module

torchrl.utils.test_schedule.exp_decay_schedule()

torchrl.utils.test_schedule.inverted_schedule()

torchrl.utils.test_schedule.schedule()

torchrl.utils.test_schedule.test_exp_decay(exp_decay_schedule: torchrl.utils.schedule.ExpDecaySchedule)

torchrl.utils.test_schedule.test_exp_decay_asymptote(exp_decay_schedule: torchrl.utils.schedule.ExpDecaySchedule)

torchrl.utils.test_schedule.test_five_steps(schedule: torchrl.utils.schedule.LinearSchedule)

torchrl.utils.test_schedule.test_overflow_steps(schedule: torchrl.utils.schedule.LinearSchedule)

torchrl.utils.test_schedule.test_underflow_steps(inverted_schedule: torchrl.utils.schedule.LinearSchedule)

torchrl.utils.test_storage module

torchrl.utils.test_storage.test_bounded(n, size)

torchrl.utils.test_storage.test_unbounded(n)

Module contents