DistributionalDQNnet

class torchrl.modules.DistributionalDQNnet(*args, **kwargs)

Distributional Deep Q-Network.

Parameters:

• DQNet (nn.Module) – (deprecated) Q-Network with output length equal to the number of atoms: output.shape = [*batch, atoms, actions].

• in_keys (list of str or tuples of str) – input keys to the log-softmax operation. Defaults to ["action_value"].

• out_keys (list of str or tuples of str) – output keys to the log-softmax operation. Defaults to ["action_value"].

forward(tensordict)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.