Adam

class maze.train.trainers.es.optimizers.adam.Adam(step_size, beta1=0.9, beta2=0.999, epsilon=1e-08)

Adam optimizer

setup(policy: TorchPolicy) → None

prepare optimizer for training