- class torchrl.modules.ObsDecoder(channels=32, num_layers=4, kernel_sizes=None, depth=None)¶
Observation decoder network.
Takes the deterministic state and the stochastic belief and decodes it into a pixel observation.
channels (int, optional) – Number of hidden units in the last layer. Defaults to 32.
num_layers (int, optional) – Depth of the network. Defaults to 4.
kernel_sizes (int or list of int, optional) – the kernel_size of each layer. Defaults to
[5, 5, 6, 6]if num_layers if 4, else
 * num_layers.
- forward(state, rnn_hidden)¶
Defines the computation performed at every call.
Should be overridden by all subclasses.
Although the recipe for forward pass needs to be defined within this function, one should call the
Moduleinstance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.