StackedBidirectionalRNN.Config¶
Component: StackedBidirectionalRNN
-
class
StackedBidirectionalRNN.
Config
[source] Bases:
Module.Config
Configuration class for StackedBidirectionalRNN.
Number of features in the hidden state of the RNN. Defaults to 32.
Type: int
-
num_layers
¶ Number of recurrent layers. Eg. setting num_layers=2 would mean stacking two RNNs together to form a stacked RNN, with the second RNN taking in the outputs of the first RNN and computing the final result. Defaults to 1.
Type: int
-
dropout
¶ Dropout probability to use. Defaults to 0.4.
Type: float
-
bidirectional
¶ If True, becomes a bidirectional RNN. Defaults to True.
Type: bool
-
rnn_type
¶ Which RNN type to use. Options: “rnn”, “lstm”, “gru”.
Type: str
-
concat_layers
¶ Whether to concatenate the outputs of each layer of stacked RNN.
Type: bool
All Attributes (including base classes)
- load_path: Optional[str] =
None
- save_path: Optional[str] =
None
- freeze: bool =
False
- shared_module_key: Optional[str] =
None
- hidden_size: int =
32
- num_layers: int =
1
- dropout: float =
0.0
- bidirectional: bool =
True
- rnn_type: RnnType =
<RnnType.LSTM: 'lstm'>
- concat_layers: bool =
True
Default JSON
{
"load_path": null,
"save_path": null,
"freeze": false,
"shared_module_key": null,
"hidden_size": 32,
"num_layers": 1,
"dropout": 0.0,
"bidirectional": true,
"rnn_type": "lstm",
"concat_layers": true
}