def __init__(self):
Model.__init__(self)
self.fc_value = self.build_network(output_dim=1)
self.fc_advantage = self.build_network(output_dim=len(config.actions))
self.optimizer_fc_value = optimizers.Adam(alpha=config.rl_learning_rate, beta1=config.rl_gradient_momentum)
self.optimizer_fc_value.setup(self.fc_value)
self.optimizer_fc_value.add_hook(optimizer.GradientClipping(10.0))
self.optimizer_fc_advantage = optimizers.Adam(alpha=config.rl_learning_rate, beta1=config.rl_gradient_momentum)
self.optimizer_fc_advantage.setup(self.fc_advantage)
self.optimizer_fc_advantage.add_hook(optimizer.GradientClipping(10.0))
self.load()
self.update_target()
评论列表
文章目录