The list of actions the model can take. (keeping the order the same is important for pretrained models)

The percentage of units to dropout between the LSTM cell layer and the dense. Useful for regularizing the model. It's disabled by default (value = 0).

The list of featurizers the model uses. (keeping the order the same is important for pretrained models)

The output size of the LSTM cell. Default is set to 32 units.

The optimization algorithm used for training. By default, Adam with a learning rate of 0.01 is used.

Temperature of the model softmax, used to calibrate confidence estimation. By default, the temperature is 1 but you usually want it higher to make less overconfident.