DTLoss

class torchrl.objectives.DTLoss(*args, **kwargs)

TorchRL implementation of the Online Decision Transformer loss.

Presented in “Decision Transformer: Reinforcement Learning via Sequence Modeling” <https://arxiv.org/abs/2106.01345>

Parameters:

actor_network (ProbabilisticActor) – stochastic actor

Keyword Arguments:

• loss_function (str) – loss function to use. Defaults to "l2".

• reduction (str, optional) – Specifies the reduction to apply to the output: "none" | "mean" | "sum". "none": no reduction will be applied, "mean": the sum of the output will be divided by the number of elements in the output, "sum": the output will be summed. Default: "mean".

forward(tensordict: TensorDictBase) → TensorDictBase

Compute the loss for the Online Decision Transformer.